CN114727110A - Method and system for data processing - Google Patents

Abstract

The data processing method and system provided in this specification can perform boundary adjustment and encoding spectrum adjustment on the initial data, so as to enhance the boundary within the preset range in the initial frame, and make the amplitude of the selected area in the initial frame smoothly Reduce, reduce the amount of data information, reduce data loss while improving the efficiency of data compression, and avoid loss of details. When decompressing the compressed data, use the parameters corresponding to the encoding spectrum adjustment to perform decoding spectrum adjustment on the compressed data, and perform boundary correction on the boundary information within the preset range to eliminate noise, and denoise the data while restoring the data. , to make the decompressed data clearer. The method and system can improve data compression efficiency, improve transmission efficiency, reduce data loss, and can improve the clarity of decompressed data.

Description

Method and system for data processing

technical field

This specification relates to the field of data processing, and in particular, to a method and system for data processing.

Background technique

With the increasing popularity of Internet technology, especially the popularization of mobile terminals, more and more types of data have emerged in communication networks. With the popularization of computers, more and more data are occupying more and more network and Store resources such as video data, audio data, and so on. Data often contains a huge amount of information, and the requirements for storage and transmission are often very high. Therefore, in order to facilitate storage and transmission, it is often necessary to compress the data, and then decompress and restore the compressed data when necessary. Therefore, data compression and decompression techniques are being used more and more.

For example, over the past few decades, video and image compression techniques have been increasingly used. Videos often contain a huge amount of information. From traditional radio, film and television to a large number of surveillance and Internet applications, compressed video and images are taking up more and more network and storage resources. This makes it occupy a lot of network resources if the original data of a piece of video is transmitted from one terminal to another terminal through the network. This makes it difficult to achieve smooth transmission of pictures in the case of some real-time video transmission. Therefore, video data must be compressed at the data compression device before being transmitted to facilitate transmission. After the compressed video is transmitted to the data decompression device through the transmission medium, the data decompression device decompresses the video to at least partially restore the video image.

The main video compression standards in the prior art are the H.264 and H.265 standards. Before transmission, an encoder is usually used to compress the video as a whole according to the H.264 and H.265 standards, and after transmission, the overall video is decompressed by a decoder according to the H.264 and H.265 standards. However, the above-mentioned processing method for overall video compression is still unsatisfactory in terms of the balance between the amount of calculation and the resolution of the video after decompression. This is because the H.264 and H.265 standards use various complex algorithms to generate the predicted frame of the original frame when processing the original video, and then record the residual between the original frame and the predicted frame . The closer the predicted frame is to the original frame, the smaller the residual, and the smaller the amount of data encoded for a piece of video. To make encoding easier, a common method is to reduce the high-frequency information in the original frame image by filtering the original frame. It can be known from the Fourier transform that the frequency information of the boundary part of the object in the picture is often rich, and the high frequency component of the boundary part is usually larger than the high frequency component of other flat areas. Therefore, although the frame image with reduced high-frequency information becomes visually blurred (that is, the definition of the image is reduced), it can make the residual between the predicted frame and the filtered original frame smaller. In this way, the amount of computation required for video encoding and the encoded data stream are greatly reduced. However, the techniques of frame prediction are very complex and consume a lot of computational resources. Taking a video encoding and decoding system as an example, the amount of computation increases by about 10 times for every 30% to 40% increase in encoding efficiency on average. At the same time, the transmitted data is decompressed and the definition is reduced, and there are often various noises, such as block effect or ringing effect. The block effect refers to that in image processing, the block-based Fourier transform causes discontinuity at the boundary of the image. The ringing effect means that in image processing, when performing spectrum adjustment processing on an image, if the selected spectrum adjustment function has a relatively rapid change in value (that is, there is an area where the derivative changes sharply), the output will be changed. The image produces grayscale oscillations where the grayscale changes sharply, just like the air oscillations produced by a clock being struck. The noise mostly occurs at the image boundary. If an output image has strong noise, it cannot meet people's increasing demands for data clarity. Therefore, how to further improve the compression efficiency of data, improve the clarity of data after decompression, and eliminate noise has always been the goal pursued in the field of data compression and decompression technology.

Therefore, in order to improve the transmission efficiency of data and the clarity after data decompression, a data processing method and system with higher compression efficiency and clearer data decompression are required.

SUMMARY OF THE INVENTION

This specification provides a method and system for data processing with higher compression efficiency and clearer data decompression. Taking video data as an example, the data processing method and system can adjust the boundaries within a small range in the initial frame in the initial video data through a small-scale gamma algorithm, so as to avoid the gap between adjacent pixels in the image. Boundary information in a small and small range is lost in the process of data compression (prediction and residual) to avoid loss of details; at the same time, the method and system can perform coding spectrum adjustment on the initial frame, so that the initial frame is The reduction of the signal strength in the selected frequency domain causes the amplitude of the selected frequency region in the initial frame to decrease smoothly, thereby reducing the amount of data information. Then, the data after spectrum adjustment is encoded (prediction and residual) to obtain compressed frames, which improves the efficiency of data compression. The encoding spectrum adjustment can reduce the amount of data information in the initial frame, and can improve the efficiency of data compression during prediction and residual error calculation. During data decompression, the method and system may perform decoded spectral adjustment and boundary correction on the compressed frame. The method and system can first decode the compressed frame, and then use the parameters corresponding to the encoding end to perform decoding spectrum adjustment on the decoded data, and the decoding spectrum adjustment can make the intermediate frequency and high frequency regions in the decoded data. The component is filtered, that is, the data that is more ambiguous than the decoded data is obtained, and the boundary information in the initial frame can be obtained by subtracting the decoded data and the data of the intermediate frequency and high frequency regions after adjustment of the decoded spectrum; then The method and system can weaken the boundary in the small range in the boundary information through the gamma algorithm in the small range, so as to perform the boundary correction to eliminate the noise in the boundary information; finally, the method and system will The decompressed frame can be obtained by superimposing the denoised boundary information and the decoded data. The decoding spectrum adjustment corresponds to the encoding spectrum adjustment, and there is a corresponding relationship between the encoding spectrum adjustment function and the decoding spectrum adjustment function, so that the compressed data adjusted by the encoding spectrum can be restored to the original frame definition or even higher. the sharpness of the initial frame. That is to say, without significantly increasing the computational complexity of encoding and decoding, the decoding end needs to at least restore the decompressed data within the important frequency to the definition of the original frame, or even obtain a definition that exceeds the original frame. Since the initial frame has only undergone signal attenuation in the frequency domain instead of filtering in the frequency domain in the important frequency region, the information in the important frequency domain is not missing, so it can be adjusted according to the encoding spectrum adjustment function and the decoding spectrum adjustment function. According to the relationship between the functions and their respective characteristics, the encoding spectrum adjustment function and the decoding spectrum adjustment function are designed to restore the information on the important frequencies in the initial frame. The method and system can significantly improve data compression efficiency, improve data transmission efficiency, reduce data loss, avoid detail loss, eliminate noise, and improve the clarity of decompressed data.

Based on this, in a first aspect, this specification provides a method for data processing, including: selecting an initial frame in initial data, where the initial frame includes initial data with a preset number of bytes; and performing data compression on the initial frame , to obtain a compressed frame, wherein the data compression includes performing boundary adjustment on the initial frame and performing encoding spectrum adjustment on the compressed frame, where the compressed frame includes the initial frame and the initial frame in the data compression During the process, it becomes any data state before the compressed frame, wherein the encoding spectrum adjustment includes convolving the under-compressed frame with an encoding convolution check, so as to smoothly reduce the intermediate frequency of the under-compressed frame in the frequency domain The magnitude of the area.

In some embodiments, the boundary adjustment includes adjusting a boundary whose boundary value is within a first preset range in the boundary of the initial frame by a first gamma algorithm whose gamma value is less than 1.

In some embodiments, the performing data compression on the initial frame includes: first performing the boundary adjustment on the initial frame, and then performing the encoding spectrum adjustment; or performing the encoding on the initial frame first spectrum adjustment, and then perform the boundary adjustment.

In some embodiments, performing the boundary adjustment on the initial frame before performing the encoding spectrum adjustment includes: performing the boundary adjustment on the initial frame to obtain a first enhanced frame; and performing the boundary adjustment on the initial frame; Performing the encoding spectrum adjustment and encoding on the first enhanced frame, including one of the following methods: first performing the encoding spectrum adjustment on the first enhanced frame, and then predicting the first enhanced frame after the encoding spectrum adjustment and residuals, wherein the compressed frame includes the first enhanced frame; the first enhanced frame is first predicted to obtain a predicted frame, and then the first enhanced frame and the predicted frame are subjected to the Coded spectrum adjustment and residual error calculation, wherein the compressed frame includes the first enhanced frame and the predicted frame; and the first enhanced initial frame is first predicted and residual error calculated, and then the residual error is calculated. The encoding spectrum adjustment is performed on the difference, wherein the under-compression frame includes the residual.

In some embodiments, performing the boundary adjustment on the initial frame to obtain the first enhanced frame includes: adjusting the initial frame through a first adjustment function to obtain the first frame, so that the initial frame The components in the low frequency region of the frame in the frequency domain are preserved and the components in the intermediate frequency to the high frequency region are attenuated; the difference between the initial frame and the first frame is obtained to obtain a first boundary, and the first boundary includes the initial frame boundary information; through the first gamma algorithm, adjust the boundary of the first boundary whose boundary value is within the first preset range to obtain an enhanced boundary; and combine the first frame with the The enhanced boundaries are superimposed to obtain the first enhanced frame.

In some embodiments, before obtaining the enhanced boundary, performing the boundary adjustment on the initial frame to obtain a first enhanced frame further includes: enhancing the first boundary by a first coefficient, wherein , the first coefficient is any number greater than 1, and the first boundary includes a boundary enhanced by the first coefficient.

In some embodiments, the adjusting the boundary of the first boundary whose boundary value is within the first preset range to obtain an enhanced boundary includes: adjusting the first boundary by using a second adjustment function , obtain a second boundary, so that the components of the low frequency region of the first boundary in the frequency domain are preserved and the components of the intermediate frequency to the high frequency region are attenuated; and through the first gamma algorithm, the second boundary is The middle boundary value is adjusted to the boundary within the first preset range to obtain the enhanced boundary.

In some embodiments, the performing the encoding spectrum adjustment on the initial frame first, and then performing the boundary adjustment, includes: performing the encoding spectrum adjustment on the initial frame, and obtaining an encoded spectrum adjustment frame; performing the boundary adjustment on the encoded spectrum adjustment frame to obtain a second enhanced frame; and performing prediction and residual error calculation on the second enhanced frame.

In some embodiments, performing the boundary adjustment on the encoded spectrum adjustment frame to obtain a second enhanced frame includes: taking a difference between the initial frame and the encoded spectrum adjustment frame to obtain a first boundary, and the The first boundary includes boundary information of the initial frame; the first gamma algorithm is used to adjust the boundary whose boundary value is within the first preset range in the first boundary to obtain an enhanced boundary; A difference between the enhancement boundary and the first boundary is obtained to obtain an adjustment value; and the second enhancement frame is obtained by superimposing the coded spectrum adjustment frame and the adjustment value.

In some embodiments, before the obtaining the enhanced boundary, performing the boundary adjustment on the encoded spectrum adjustment frame to obtain a second enhanced frame, further comprising: enhancing the first boundary by a first coefficient , wherein the first coefficient is any number greater than 1, and the first boundary includes a boundary enhanced by the first coefficient.

In some embodiments, the adjusting the boundary of the first boundary whose boundary value is within the first preset range to obtain an enhanced boundary includes: adjusting the first boundary by using a second adjustment function , obtain a second boundary, so that the components of the low frequency region of the first boundary in the frequency domain are preserved and the components of the intermediate frequency to the high frequency region are attenuated; and through the first gamma algorithm, the second boundary is The middle boundary value is adjusted to the boundary within the first preset range to obtain the enhanced boundary.

In some embodiments, the performing encoding spectrum adjustment on the under-compressed frame includes: determining a frame type of the initial frame, where the frame type includes an intra-frame prediction frame, a forward-prediction frame, and a bidirectionally-predicted frame. at least one; and, based on the frame type of the initial frame, selecting a convolution kernel from the encoding convolution kernel group as the encoding convolution kernel to perform convolution on the compressed frame.

In some embodiments, performing the convolution on the on-press frame includes: performing convolution on the on-press frame in at least one of a vertical direction, a horizontal direction, and an oblique direction.

In some embodiments, the encoding spectral adjustment is such that the amplitude in the framed high frequency region is smoothly reduced in the frequency domain.

In some embodiments, the encoding spectral adjustment causes a smooth reduction in the amplitude of the low frequency region of the framed frame in the frequency domain, and the encoding spectral adjustment reduces the amplitude of the low frequency region of the framed frame. The magnitude is lower than the magnitude of the magnitude reduction in the mid-frequency region.

In some embodiments, the amplitude adjustment gain of the encoding spectrum adjustment at any frequency in the frequency domain of the frame is greater than zero.

In a second aspect, this specification also provides a data processing system, comprising: at least one storage medium and at least one processor, wherein the at least one storage medium stores at least one instruction set for data processing; the at least one processing The processor is communicatively connected to the at least one storage medium, wherein, when the system is running, the at least one processor reads the at least one instruction set, and executes the first instruction set of the present specification according to the instructions of the at least one instruction set. The method of data processing described in the aspect.

In a third aspect, this specification also provides a data processing method, including: acquiring compressed data, the compressed data including a compressed frame obtained by data compression of an initial frame; and decompressing the compressed frame to obtain a decompressed frame , the data decompression includes performing decoding spectrum adjustment and boundary correction on the decompressed frame, where the deframe includes the compressed frame and the compressed frame becomes any data before the decompressed frame during the data decompression process state, wherein the decoded spectral adjustment includes convolving the deframed with a decoded convolution kernel so that the deframed amplitude in the frequency domain is smoothly reduced to filter components in the mid- to high-frequency region, so There is a preset correlation relationship between the encoding spectrum adjustment and the decoding spectrum adjustment, and the boundary correction includes performing a second gamma algorithm on the boundary where the boundary value is within the second preset range in the deframed boundary. Correction for noise reduction.

In some embodiments, the second gamma algorithm includes a gamma algorithm with a gamma value greater than 1, so as to weaken the boundary of the de-framed boundary whose boundary value is within the second preset range.

In some embodiments, the performing data decompression on the compressed frame includes: performing the decoding spectrum adjustment on the decompressed frame to obtain a decoded spectrum adjustment frame; adjusting the decompression frame and the decoded spectrum The frame difference is obtained to obtain a third boundary, and the third boundary is the boundary of the frame being decoded, including the boundary information of the initial frame; through the second gamma algorithm, the boundary value in the third boundary is calculated. The boundary within the second preset range is attenuated to denoise the third boundary to obtain a noise reduction boundary; and the decompressed frame is obtained by superimposing the noise reduction boundary and the decompressed frame.

In some embodiments, before the denoising boundary is obtained, performing data decompression on the compressed frame includes: enhancing the third boundary by a second coefficient, wherein the second coefficient is greater than Any number of 1, the third boundary includes the boundary enhanced by the second coefficient.

In some embodiments, the second gamma algorithm is used to attenuate the boundary whose boundary value is within the second preset range in the third boundary, so as to denoise the third boundary , to obtain a noise reduction boundary, including: adjusting the third boundary through a third adjustment function to obtain a fourth boundary, so that the component of the third boundary in the low frequency region in the frequency domain is retained and the intermediate frequency to high frequency region is The component of is filtered; and the boundary whose boundary value is within the second preset range of the fourth boundary is weakened by the second gamma algorithm.

In some embodiments, before performing the decoding spectrum adjustment on the decoded frame to obtain a decoded spectrum adjustment frame, performing data decompression on the compressed frame further includes: decoding the compressed frame , to obtain a decoded frame, and the decoded frame includes the decoded frame.

In some embodiments, the data compression includes coded spectral adjustment, including convolving the compressed frame with an coded convolution kernel to smoothly reduce the amplitude of the compressed frame in the intermediate frequency region in the frequency domain, the The compressed frame includes the initial frame and any data state before the initial frame becomes the compressed frame in the data compression process, wherein the decoding convolution kernel corresponds to the encoding convolution kernel.

In a fourth aspect, this specification also provides a data processing system, comprising at least one storage medium and at least one processor, the at least one storage medium stores at least one instruction set for data processing; and the at least one processing The processor is communicatively connected to the at least one storage medium, wherein when the system is running, the at least one processor reads the at least one instruction set and executes the third aspect of the present specification according to the instructions of the at least one instruction set The described data processing method.

Other functions of the data processing method and system provided by this specification will be partially listed in the following description. From the description, what is presented in the following figures and examples will be apparent to those of ordinary skill in the art. The inventive aspects of the data processing methods, systems, and storage media provided by this specification can be fully explained by practice or use of the methods, apparatus, and combinations described in the detailed examples below.

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present specification more clearly, the following briefly introduces the drawings that are used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present specification. For those of ordinary skill in the art, other drawings can also be obtained from these drawings without creative effort.

FIG. 1 shows a schematic diagram of a data processing system provided according to an embodiment of the present specification;

FIG. 2 shows a schematic diagram of a data compression device for data processing provided according to an embodiment of the present specification;

3A shows a flowchart of data compression and data decompression provided according to an embodiment of the present specification;

FIG. 3B shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification;

3C shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification;

FIG. 3D shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification;

4A shows a flowchart of a data processing method for compressing data provided according to an embodiment of the present specification;

4B shows a flowchart of a data processing method for compressing data provided according to an embodiment of the present specification;

4C shows a flowchart of a data processing method for compressing data provided according to an embodiment of the present specification;

4D shows a flowchart of a data processing method for compressing data provided according to an embodiment of the present specification;

FIG. 5A shows a flowchart of acquiring a first enhanced frame according to an embodiment of the present specification;

FIG. 5B shows a flowchart of performing boundary adjustment according to an embodiment of the present specification;

FIG. 5C shows another flowchart of boundary adjustment provided according to an embodiment of the present specification;

FIG. 6 shows a graph for obtaining a second adjustment function according to an embodiment of the present specification;

FIG. 7 shows a graph of a gamma algorithm provided according to an embodiment of the present specification;

8A shows a graph of an encoding spectrum adjustment function provided according to an embodiment of the present specification;

FIG. 8B shows a graph of an encoding spectrum adjustment function provided according to an embodiment of the present specification;

FIG. 9 shows a flowchart of acquiring a second enhanced frame according to an embodiment of the present specification;

10 shows a flowchart of a method for data processing for decompressing a compressed frame provided according to an embodiment of the present specification;

11A shows a flowchart of decoding spectrum adjustment and boundary correction provided according to an embodiment of the present specification;

FIG. 11B shows a flowchart of boundary correction according to an embodiment of the present specification;

FIG. 11C shows another flowchart of performing boundary correction according to an embodiment of the present specification;

FIG. 12A shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification;

12B shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification;

FIG. 12C shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification;

12D shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification;

FIG. 12E shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification;

13A shows a graph of the overall adjustment function H ₀ (f), the encoding spectrum adjustment function H ₁ (f) and the decoding spectrum adjustment function H ₂ (f) of a normal mode provided according to an embodiment of the present specification;

13B shows a graph of the overall adjustment function H ₀ (f), the encoding spectrum adjustment function H ₁ (f) and the decoding spectrum adjustment function H ₂ (f) of an enhancement mode provided according to an embodiment of the present specification;

FIG. 14A shows an example diagram of a decompressed image without boundary correction provided according to an embodiment of the present specification; and

FIG. 14B shows an example diagram of a decompressed image with boundary correction according to an embodiment of the present specification.

Detailed ways

The following description provides specific application scenarios and requirements of this specification, and is intended to enable those skilled in the art to make and use the content of this specification. Various partial modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments and without departing from the spirit and scope of the description. application. Thus, this specification is not to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the claims.

The terminology used herein is for the purpose of describing particular example embodiments only and is not limiting. For example, as used herein, the singular forms "a," "an," and "the" can include the plural forms as well, unless the context clearly dictates otherwise. When used in this specification, the terms "comprising", "comprising" and/or "comprising" are meant to refer to the associated integer, step, operation, element and/or component being present, but not excluding one or more other features , integers, steps, operations, elements, components and/or groups exist or other features, integers, steps, operations, elements, components and/or groups may be added to the system/method.

These and other features of this specification, as well as the operation and function of related elements of structure, and the economics of assembly and manufacture of parts, may be significantly improved in view of the following description. Reference is made to the accompanying drawings, all of which form a part of this specification. However, it should be clearly understood that the drawings are for illustration and description purposes only and are not intended to limit the scope of the present specification. It should also be understood that the figures are not drawn to scale.

The flowcharts used in this specification illustrate the operation of a system implementation according to some embodiments in this specification. It should be clearly understood that the operations of the flowcharts may be implemented out of sequence. Instead, operations may be implemented in reverse order or simultaneously. Additionally, one or more other operations can be added to the flowchart. One or more operations can be removed from the flowchart.

In one aspect, the present specification provides a data processing system 100 (hereinafter referred to as the system 100). In the second aspect, this specification describes a data processing method P200 for compressing data, and in the third aspect, this specification describes a data processing method P300 for decompressing a compressed frame. The data processing methods P200, P300 and the system 100 can be used for data compression and decompression, so as to improve the transmission efficiency of the data and save resources and space. The data may be non-real-time data or real-time data. From traditional radio, film and television to a large number of surveillance and Internet applications, there are all kinds of data. For example, the data may be non-real-time video data, audio data, or image data, and the like. The data may also be real-time map data, real-time sensor data, real-time video surveillance data, network monitoring data, meteorological data, aerospace data, and the like. For example, the data may be map data received from a base station by an autonomous vehicle during driving. This specification does not limit the specific categories of the data. The data processing methods P200, P300 and the system 100 described in this specification are consistent in the methods and steps taken when processing different types of data. For the convenience of presentation, this specification will take the processing of video data as an example. describe.

In data compression and data decompression, compression and decompression are often performed in units of frames. A frame is a processing unit that makes up a sequence of data. One or more initial frames may be included in the initial data. Each initial frame includes a preset number of bytes of initial data. In video compression, the initial data may be initial video data, and the initial frame may be a frame image in the initial video data. In traditional video compression technologies, H.264 and H.265 standards are usually used to encode initial video data, so as to achieve the purpose of compressing the video data. The main technical means used to encode video data in the H.264 and H.265 standards is predictive encoding, that is, predicting the initial data in the video data to obtain a predicted value, and then comparing the predicted value with the initial value of the initial data. The residual value is subtracted to compress the video data. During recovery and decompression (ie, decoding), the initial frame can be recovered by adding the residual and predicted values.

The data processing methods P200, P300 and system 100 provided in this specification can combine encoding spectrum adjustment and encoding during data compression, so as to reduce the amount of data during encoding, improve the compression efficiency of video data, and improve the transmission efficiency of video; During data decompression, decoding spectrum adjustment and decoding can be combined to decompress the compressed data after encoding spectrum adjustment and encoding, so that the decompressed data can be restored to the original data.

FIG. 1 shows a schematic diagram of a system 100 for data processing. System 100 may include data compression device 200 , data decompression device 300 , and transmission medium 120 .

The data compression device 200 may receive an initial frame in the initial data to be compressed, and use the data processing method P200 proposed in this specification to compress the initial data to generate a compressed frame. The data compression apparatus 200 may store data or instructions for performing the data processing method P200 described in this specification, and execute the data and/or instructions.

The data processing method P200 may perform data compression on the video data. The data compression may be boundary adjustment of initial frames in the video data and encoding spectrum adjustment and encoding of the video data. Specifically, the data processing method P200 may perform the boundary adjustment on the boundary of a small range in the video data by using a gamma algorithm to adjust the boundary of the small range, so as to avoid a boundary with a small difference in pixel values between adjacent pixels. is lost in the encoding process to avoid loss of details; at the same time, the data processing method P200 can also compress the video data by combining the encoding spectrum adjustment and encoding to obtain compressed frames, so as to further improve the compression ratio of the video data, Improve the efficiency of video transmission. Specifically, the data processing method P200 may use a gamma algorithm with a gamma value less than 1 to perform the boundary adjustment on the video data. The encoding spectrum adjustment refers to adjusting the amplitude of the spectrogram of the data to be processed. For example, the encoding spectrum adjustment may attenuate the amplitude of the data to be processed in the frequency domain, thereby reducing the amount of information in the data to be processed, such as attenuating the selected frequency of the data to be processed in the frequency domain The amplitude of the region, such as the amplitude of the medium frequency region, the amplitude of the high frequency region, such as the amplitude of the low frequency to the medium frequency region, and the amplitude of the medium frequency to the high frequency region, and so on. It can be understood by those skilled in the art that the frequency components in the selected frequency region of the coded spectrum adjusted data become smaller, and the amount of information in the data is reduced. Therefore, the coding efficiency of the coded spectrum adjusted data can be improved. Increase, increase the compression ratio.

The data decompression device 300 may receive the compressed frame, and use the data processing method P300 proposed in this specification to decompress the compressed frame to obtain the decompressed frame. The data decompression device 300 may store data or instructions for performing the data processing method P300 described in this specification, and execute the data and/or instructions.

The data processing method P300 may perform data decompression on the compressed frame subjected to the data compression by the data processing method P200 to restore the video data. The data decompression may be a decoding spectral adjustment and boundary correction of the compressed frame. The data processing method P300 may use a combination of decoding (that is, restoring the compressed frame according to the residual value and the predicted value) and decoding spectrum adjustment to decompress the compressed frame to restore the data in the compressed frame. The data processing method P300 can perform decoding spectrum adjustment on the compressed data through the decoding spectrum adjustment function; difference between the compressed data and the decoded data to obtain the boundary information of the initial frame; and perform the boundary information on the boundary information through a gamma algorithm. The decompression frame can be obtained by superimposing the boundary information after noise removal and the decoded data, and the boundary correction can Noise information is removed to make the decompressed frame clearer. Specifically, the decoding spectrum adjustment uses a smooth transition low-pass filter to filter the components in the intermediate frequency and high frequency regions in the decoded data. Therefore, the decoded data can effectively avoid ringing effects, thereby enabling decompression The latter data is clearer. The decoding spectrum adjustment can make the data subjected to the encoding spectrum adjustment completely or approximately restored to the state before the encoding spectrum adjustment without considering other calculation errors, or even beyond the state before the encoding spectrum adjustment.

Therefore, the data processing methods P200, P300 and the system 100 can significantly improve the compression efficiency of video data, reduce the data loss during video data compression, improve the transmission efficiency of the video, the restoration rate and the clarity of the decompressed video, and reduce the content of the decompressed video. noise. The specific processes of the encoding spectrum adjustment and the boundary correction and the decoding spectrum adjustment and the boundary correction will be described in detail in the following description.

Data compression apparatus 200 and data decompression apparatus 300 may include a wide range of devices. For example, data compression apparatus 200 and data decompression apparatus 300 may include desktop computers, mobile computing devices, notebook (eg, laptop) computers, tablet computers, set-top boxes, handsets such as smartphones, televisions, cameras, display devices, digital media Players, video game consoles, in-vehicle computers, or the like.

As shown in FIG. 1 , the data compression device 200 and the data decompression device 300 may be connected through the transmission medium 120 . Transmission medium 120 may facilitate the transfer of information and/or data. Transmission medium 120 may be any data carrier that can transmit compressed frames from data compression device 200 to data decompression device 300 . For example, transmission medium 120 may be a storage medium (eg, an optical disk), a wired or wireless communication medium. The communication medium may be a network. In some embodiments, the transmission medium 120 may be any type of wired or wireless network, or a combination thereof. For example, the transmission medium 120 may include a cable network, a wired network, a fiber optic network, a telecommunication network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), Wide Area Network (WAN), Public Switched Telephone Network (PSTN), Bluetooth network, ZigBee network, Near Field Communication (NFC) network or similar network. One or more components of data decompression device 300 and data compression device 200 may be connected to transmission medium 120 to transmit data and/or information. Transmission medium 120 may include routers, switches, base stations, or other devices that facilitate communication from data compression device 200 to data decompression device 300 . In other embodiments, the transmission medium 120 may be a storage medium such as mass storage, removable storage, volatile read-write memory, read-only memory (ROM), or the like, or any combination thereof. Exemplary mass storage may include non-transitory storage media such as magnetic disks, optical disks, solid state drives, and the like. Removable storage may include flash drives, floppy disks, optical disks, memory cards, zip disks, tapes, etc. Typical volatile read-write memory may include random access memory (RAM). RAM may include Dynamic RAM (DRAM), Double Date Rate Synchronous Dynamic RAM (DDRSDRAM), Static RAM (SRAM), Thyristor RAM (T-RAM), and Zero Capacitance RAM (Z-RAM), among others. ROM may include masked ROM (MROM), programmable ROM (PROM), virtual programmable ROM (PEROM), electronically programmable ROM (EEPROM), compact disk (CD-ROM), digital versatile disk ROM, and the like. In some embodiments, the transmission medium 120 may be a cloud platform. Just as an example, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an inter-cloud cloud, etc., or a form similar to the above-mentioned forms, or any combination of the above-mentioned forms.

As shown in FIG. 1, the data compression device 200 receives the initial data, and executes the instructions of the data processing method P200 described in this specification, compresses the initial data, and generates a compressed frame; the compressed frame is transmitted to the data through the transmission medium 120. Decompression device 300; the data decompression device 300 executes the instructions of the data processing method P300 described in this specification, and decompresses the data of the compressed frame to obtain the decompressed frame.

FIG. 2 shows a schematic diagram of a data compression device 200 for data processing. The data compression device 200 may perform the method P200 of data processing described in this specification. The data processing method P200 is described elsewhere in this specification. For example, the data processing method P200 is introduced in the description of FIGS. 4A to 9 .

As shown in FIG. 2 , the data compression apparatus 200 includes at least one storage medium 230 and at least one compression end processor 220 . In some embodiments, the data compression device 200 may also include a communication port 250 and an internal communication bus 210 . Meanwhile, the data compression apparatus 200 may further include an I/O component 260 .

The internal communication bus 210 may connect various system components, including the storage medium 230 and the compressor side processor 220 .

I/O component 260 supports input/output between data compression device 200 and other components.

The storage medium 230 may include data storage devices. The data storage device may be a non-transitory storage medium or a temporary storage medium. For example, the data storage device may include one or more of a magnetic disk 232 , a read only storage medium (ROM) 234 or a random access storage medium (RAM) 236 . Storage medium 230 also includes at least one set of instructions stored in the data storage device. The instructions are computer program code, which may include programs, routines, objects, components, data structures, procedures, modules, etc. that perform the methods of data processing provided by this specification.

The communication port 250 is used for data communication between the data compression device 200 and the outside world. For example, the data compression device 200 may be connected to the transmission medium 120 through the communication port 250 .

At least one compression end processor 220 is communicatively connected with at least one storage medium 230 through an internal communication bus 210 . At least one compressor side processor 220 is used for executing the above at least one instruction set. When the system 100 is running, the at least one compression end processor 220 reads the at least one instruction set, and executes the data processing method P200 according to the instructions of the at least one instruction set. The compressor end processor 220 may perform all steps included in the data processing method P200. Compressor side processor 220 may be in the form of one or more processors, and in some embodiments, compressor side processor 220 may comprise one or more hardware processors, such as microcontrollers, microprocessors, reduced instruction set computers (RISC), Application Specific Integrated Circuit (ASIC), Application Specific Instruction Set Processor (ASIP), Central Processing Unit (CPU), Graphics Processing Unit (GPU), Physical Processing Unit (PPU), Microcontroller Unit, Digital Signal Processor (DSP), Field Programmable Gate Array (FPGA), Advanced RISC Machine (ARM), Programmable Logic Device (PLD), any circuit or processor capable of performing one or more functions, etc., or any combination thereof . Just to illustrate the problem, only one compression end processor 220 is described in the data compression apparatus 200 in this specification. However, it should be noted that the data compression apparatus 200 in this specification may also include multiple processors, and thus, the operations and/or method steps disclosed in this specification may be performed by one processor as described in this specification, or may be performed by multiple processors. The processors are executed jointly. For example, if the compressor side processor 220 of the data compression device 200 performs step A and step B in this specification, it should be understood that step A and step B may also be performed jointly or separately by two different compressor side processors 220 (for example, , the first processor executes step A, the second processor executes step B, or the first and second processors jointly execute steps A and B).

Although the above structure describes the data compression apparatus 200, this structure is also applicable to the data decompression apparatus 300. The data decompression device 300 may perform the data processing method P300 described in this specification. The data processing method P300 is described elsewhere in this specification. For example, the data processing method P300 is introduced in the description of FIGS. 10 to 14B .

When the system 100 performs data compression on the video data, the encoding spectrum adjustment and the encoding sequence may be interchanged, or may be performed in an interleaved manner. The boundary adjustment may be performed before the encoding spectrum adjustment or after the encoding spectrum adjustment. Likewise, when the system 100 performs data decompression on the compressed frame, the decoding spectrum adjustment and the decoding sequence may be interchanged, or may be performed interleaved. It should be noted that, in order to ensure that the decompressed data information can restore the information in the original data, the decoding spectrum adjustment and the decoding sequence in the data decompression are the same as the encoding spectrum adjustment and the decoding sequence in the data compression. The encoding sequence should be corresponding, that is, the decoding spectrum adjustment and the decoding can be symmetrically reversed to the encoding spectrum adjustment and the encoding. For example, if the compressed frame is obtained by performing the encoding spectrum adjustment first and then performing the encoding, the compressed frame should be decoded first and then the decoded spectrum adjustment should be performed during data decompression. For the convenience of description, we define the data in the initial frame before data compression processing as P ₀ , define the encoding spectrum adjustment function corresponding to encoding spectrum adjustment as H ₁ (f), and decompress the data obtained by the data decompression device 300. The data in the decompressed frame is defined as P ₄ , and the decoding spectrum adjustment function corresponding to the decoding spectrum adjustment is defined as H ₂ (f).

In the data processing method P200, when the data compression device 200 performs data compression on the initial frame, the boundary adjustment may be performed on the initial frame first, and then the encoding spectrum adjustment is performed; The encoding spectrum adjustment is performed on the initial frame, and then the boundary adjustment is performed. 3A to 3D illustrate some flowcharts of data compression and data decompression provided according to embodiments of the present specification. In the flowcharts of data compression and data decompression shown in FIGS. 3A to 3C , the data compression device 200 first performs the boundary adjustment on the initial frame, and then performs the encoding spectrum adjustment. In the flowchart of data compression and data decompression shown in FIG. 3D , the data compression device 200 first performs the encoding spectrum adjustment on the initial frame, and then performs the boundary adjustment.

FIG. 3A shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification. As shown in FIG. 3A , the data compression device 200 performs data compression on the initial data may be: the data compression device 200 first performs the boundary adjustment on the boundary of the initial frame P ₀ to obtain the first enhanced frame. For the convenience of description, we define the data in the first enhanced frame as P′ ₀ ; and then perform the encoding spectrum adjustment and encoding on the first enhanced frame P′ ₀ . Wherein, performing the encoding spectrum adjustment and encoding on the first enhanced frame P′ ₀ may be: the data compression device 200 uses the encoding spectrum adjustment function H ₁ (f) to perform the encoding spectrum adjustment on the first enhanced frame P′ _0. Perform the encoding spectrum adjustment first, and then perform the encoding on the first enhanced frame P′ ₀ after the encoding spectrum adjustment, that is, perform prediction and residual error on the first enhanced frame P′ ₀ after the encoding spectrum adjustment, to obtain The predicted data PI and the residual data R are input into the code stream generation module for synthesis to obtain the compressed frame. For the convenience of presentation, we define the data obtained after the encoding spectrum adjustment is performed by the encoding spectrum adjustment function H ₁ (f) as P ₁ . The content of the boundary adjustment and the encoding spectrum adjustment will be introduced in detail in the following description. The data compression method shown in FIG. 3A can improve coding efficiency, further reduce the amount of data in the compressed frame, improve the compression ratio, reduce data loss, and avoid loss of details.

As shown in FIG. 3A , the data decompression device 300 performs data decompression on the compressed frame: the data decompression device 300 first performs the decoding on the compressed frame, that is, parses the compressed frame based on the code stream parsing module, Generate the predicted data PI and the residual data R; and then perform prediction according to the predicted data PI to obtain a predicted frame, which is superimposed with the residual data R to obtain a decoded frame. For the convenience of description, we define the data in the decoded frame as P ₂ . Then use the decoded spectrum adjustment function H ₂ (f) to perform the decoded spectrum adjustment and the boundary correction on the decoded frame P ₂ , and superimpose the decoded frame and the boundary-corrected data to obtain the Decompress frame _P4 for output. The specific content of the decoding spectrum adjustment and the boundary correction will be introduced in detail in the following description.

For convenience of presentation, we define the transfer function between the decompressed frame P ₄ and the original data P ₀ as the overall spectral adjustment function H ₀ (f). The method shown in FIG. 3A can reduce the amount of data in the compressed frame, thereby improving the compression ratio and coding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding detail loss.

The data compression device 200 performs data compression on the initial data may also include: incorporating the encoding spectrum adjustment into the encoding process. The encoding spectral adjustment can be performed at any stage in the encoding process. Correspondingly, the decoding spectrum adjustment can also be performed at a corresponding stage of the decoding process.

FIG. 3B shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification. As shown in FIG. 3B , the data compression device 200 performs data compression on the initial data: the data compression device 200 first performs the boundary adjustment on the boundary of the initial frame P ₀ to obtain the first enhanced frame P′ ₀ ; The first enhancement frame P' ₀ performs the encoding spectral adjustment and the encoding. Wherein, performing the encoding spectrum adjustment and the encoding on the first enhanced frame P′ ₀ may be: the data compression device 200 predicts the first enhanced frame P′ ₀ to obtain a predicted frame and predicted data PI, Then use the encoding spectrum adjustment function H 1 (f) to adjust the encoding spectrum for the predicted frame and the first enhanced frame P′ ₀ respectively, and then obtain the residual to obtain the residual data R, and then use the encoding spectrum adjustment function H ₁ (f). The data PI and the residual data R are input to the code stream generation module for synthesis to generate the compressed frame. The specific operation of data compression shown in FIG. 3B is the same as that shown in FIG. 3A, but the operation sequence is different. The content of the boundary adjustment and the encoding spectrum adjustment will be introduced in detail in the following description.

As shown in FIG. 3B , the data decompression device 300 performs data decompression on the compressed frame: the data decompression device 300 parses the compressed frame based on a code stream parsing module, and generates the prediction data PI and the residual data R ₁ ; use the decoded spectrum adjustment function H ₂ (f) to perform the decoded spectrum adjustment on the residual data R ₁ , and calculate the difference between the residual data R ₁ and the data subjected to the decoded spectrum adjustment, Obtain the boundary of the residual data R ₁ , superimpose the residual data R ₁ with the differenced data (the boundary of the residual data R ₁ ) to obtain the residual data R; then according to the predicted data PI performs prediction to obtain a predicted frame, and superimposes it with the residual data R to obtain a superimposed frame; then perform the boundary correction on the boundary in the superimposed frame, and use the data after the boundary correction as the decompression Frame _P4 is output. For the convenience of description, we define the data in the superimposed frame as P ₃ . Specifically, the specific process of performing the boundary correction on the boundary in the superimposed frame _P3 will be described in detail in the following content.

The method shown in FIG. 3B can reduce the amount of data in the compressed frame, thereby improving the compression ratio and coding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding detail loss.

FIG. 3C shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification. As shown in FIG. 3C , the data compression device 200 performs data compression on the initial data: the data compression device 200 first performs the boundary adjustment on the boundary of the initial frame P ₀ to obtain the first enhanced frame P′ ₀ ; The first enhancement frame P' ₀ performs the encoding spectral adjustment and the encoding. Wherein, performing the encoding spectrum adjustment and the encoding on the first enhanced frame P' ₀ may be: the data compression device 200 first performs the encoding on the first enhanced frame P' ₀ , that is, prediction and residual The predicted data PI and the residual data R are obtained, and then the encoded spectrum adjustment function H ₁ (f) is used for the residual data R to perform the encoding spectrum adjustment; _The data R1 and the predicted data PI are input into the code stream generation module for synthesis to generate the compressed frame. The specific operation of the data compression method shown in FIG. 3C is the same as that of the method shown in FIG. 3A, but the operation sequence is different. The content of the boundary adjustment and the encoding spectrum adjustment will be introduced in detail in the following description.

As shown in FIG. 3C , the data decompression device 300 performs data decompression on the compressed frame: the data decompression device 300 parses the compressed frame based on a code stream parsing module, and generates the prediction data PI and the residual data R ₁ ; use the decoded spectrum adjustment function H ₂ (f) to perform the decoded spectrum adjustment on the residual data R ₁ , and calculate the difference between the residual data R ₁ and the data subjected to the decoded spectrum adjustment , obtain the boundary of the residual data R ₁ , and superimpose the residual data R ₁ with the differenced data (the boundary of the residual data R ₁ ) to obtain the residual data R; then according to the prediction The data PI is predicted to obtain a predicted frame, and is superimposed with the residual data R to obtain a superimposed frame P ₃ ; then the boundary correction is performed on the boundary in the superimposed frame, and the data after the boundary correction is used as _The decompressed frame P4 is output. Specifically, the specific process of performing the boundary correction on the boundary in the superimposed frame _P3 will be described in detail in the following content.

The method shown in FIG. 3C can reduce the amount of data in the compressed frame, thereby improving the compression ratio and coding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding detail loss.

FIG. 3D shows a flow chart of data compression and data decompression provided according to an embodiment of the present specification. As shown in FIG. 3D , the data compression device 200 performs data compression on the initial data may be: the data compression device 200 first uses the encoding spectrum adjustment function H ₁ (f) to perform the encoding spectrum adjustment on the initial frame P ₀ , Then perform the boundary adjustment to obtain the data P'₁; and then perform the encoding on the data P' ₁ , that is, perform prediction on the data P' ₁ and obtain the residual, to obtain the predicted data PI and the residual data R. The prediction data PI and the residual data R are input into the code stream generation module for synthesis to obtain the compressed frame. The content of the boundary adjustment and the encoding spectrum adjustment will be introduced in detail in the following description. The data compression method shown in FIG. 3D can improve coding efficiency, further reduce the amount of data in the compressed frame, improve the compression ratio, reduce data loss, and avoid loss of details.

As shown in FIG. 3D , the data decompression device 300 performs data decompression on the compressed frame: the data decompression device 300 first performs the decoding on the compressed frame, that is, parses the compressed frame based on the code stream parsing module, Generate the predicted data PI and the residual data R; and then perform prediction according to the predicted data PI to obtain a predicted frame, which is superimposed with the residual data R to obtain a decoded frame. For the convenience of description, we define the data in the decoded frame as P ₂ . Then use the decoded spectrum adjustment function H ₂ (f) to perform the decoded spectrum adjustment and the boundary correction on the decoded frame P ₂ , and superimpose the decoded frame and the boundary-corrected data to obtain the Decompress frame _P4 for output. The specific content of the decoding spectrum adjustment and the boundary correction will be introduced in detail in the following description.

The method shown in FIG. 3D can reduce the amount of data in the compressed frame, thereby improving the compression ratio and coding efficiency of the initial data, improving the transmission efficiency of the initial data, and at the same time reducing data loss and avoiding detail loss. The specific process of the data compression and data decompression will be described in detail in the following content.

4A to 4D show flowcharts of some data processing methods P200 for compressing data provided according to embodiments of the present specification. As previously described, the data compression apparatus 200 may perform the data processing method P200. Specifically, the storage medium in the data compression device 200 may store at least one set of instruction sets. The instruction set is configured to instruct the compression processor 220 in the data compression apparatus 200 to complete the data processing method P200. When the data compression apparatus 200 is running, the compression processor 220 may read the instruction set and execute the data processing method P200.

The data processing method P200 may be the data processing method PA200 shown in FIG. 4A, the data processing method PB200 shown in FIG. 4B, the data processing method PC200 shown in FIG. 4C, or the data processing method PC200 shown in FIG. 4D. Data processing method PD200 shown. The data processing method PA200 shown in FIG. 4A corresponds to the flowchart shown in FIG. 3A . The data processing method PB200 shown in FIG. 4B corresponds to the flowchart shown in FIG. 3B . The data processing method PC200 shown in FIG. 4C corresponds to the flowchart shown in FIG. 3C . The data processing method PD200 shown in FIG. 4D corresponds to the flowchart shown in FIG. 3D .

As shown in FIG. 4A, the method PA200 may include:

SA220: Select the initial frame in the initial data.

A frame is a processing unit that makes up a sequence of data. In data processing, calculations are often performed in units of frames. The initial data may include one or more initial frames. The initial frame includes initial data of a preset number of bytes. As mentioned above, video data is used as an example for description in this specification. Therefore, the initial data may be initial video data, and the initial frame may be a frame image in the initial video data. In step SA220, the data compression device 200 may select a part of frame images from the initial data as the initial frame, or may select all frame images in the initial data as the initial frame. The data compression device 200 may select the initial frame according to the initial data application scenario. If the initial data is used in a scene that does not require high precision and compression quality, part of the frame image can be selected as the initial frame. Most of the frame images of the monitoring images are the same, and the data compression device 200 may select some frame images as the initial frame for compression and transmission. For another example, for a high-definition TV broadcast video, in order to ensure a movie viewing effect, the data compression device 200 may select all frame images as the initial frame for compression and transmission.

SA240: Perform the data compression on the initial frame to obtain a compressed frame.

The data compression may include performing the boundary adjustment on the initial frame and performing the encoding spectrum adjustment on the compressed frame. The performing the boundary adjustment on the initial frame may be to adjust a boundary whose boundary value is within a first preset range in the boundary of the initial frame by using a first gamma algorithm.

The performing the encoding spectrum adjustment on the currently-compressed frame may be inputting the currently-compressed frame into an encoding spectrum adjuster to perform encoding spectrum adjustment. The encoding spectrum adjustment refers to adjusting the amplitude of the spectrogram in the frame being compressed. For example, the encoding spectrum adjustment can be done by an attenuator. The attenuator may attenuate the amplitude of the compressed frame in the frequency domain, thereby reducing the amount of data information in the compressed frame.

The attenuator may be configured to reduce the amplitude of the selected region of the frame in its frequency domain, such as the amplitude of the mid-frequency region, the amplitude of the high-frequency region, and the amplitude of the low-to-medium frequency region, Another example, the amplitude of the mid-frequency to high-frequency region, and so on. For different forms of data, receivers have different degrees of sensitivity to frequency, so the data compression operation may select different regions in the frequency domain to perform amplitude attenuation according to different forms of data. As mentioned above, taking video data as an example, since the edge part of the object in the picture is rich in medium frequency and high frequency information, and the medium frequency and high frequency area will carry more data, so reduce the amplitude of the medium frequency to the high frequency area from the visual The above will blur the boundary data of the frame, and at the same time, the amount of information in the image will be greatly reduced. It should be noted that reducing the amplitude of the low frequency region will also reduce the amount of information in the image. Those of ordinary skill in the art can understand that, compared with the case without the encoding spectrum adjustment process, the frequency components in the low frequency to high frequency region in the intermediate state frame after the encoding spectrum adjustment process are reduced, and the amount of data information is also reduced. Therefore, the intermediate state frame processed by the encoding spectrum adjustment will have a higher compression ratio in the encoding. Different types of data can define the low, mid, and high frequency regions differently. In some embodiments, the high frequency may include frequencies between (0.33, 0.5] in the normalized frequency domain. For example, the high frequency may include 0.33, 0.34, The interval between any two frequencies of 0.35, 0.36, 0.37, 0.38, 0.39, 0.4, 0.41, 0.42, 0.43, 0.44, 0.45, 0.46, 0.47, 0.48, 0.49, 0.5, where 0.5 is the normalized maximum frequency.

Taking video data compression as an example, the data processing method P200 can compress the initial frame by using a combination of coding spectrum adjustment and coding, so that the amplitude of the intermediate frequency region is steadily reduced, so as to reduce the amount of data information and further improve the video data. The compression ratio can improve the efficiency of video transmission. The compressed frame may include the initial frame and any data state of the initial frame before the compressed frame during the data compression process, for example, the initial frame is undergoing the encoding spectrum adjustment and encoding. Any data state in the process, for example, initial frame, predicted frame, residual frame, and so on. In FIGS. 3A and 4A , the under-pressing frame may be the initial frame.

Step SA240 may be to perform the boundary adjustment on the initial frame first, and then perform the encoding spectrum adjustment. Specifically, step SA240 may include:

SA242: Perform the boundary adjustment on the initial frame P ₀ to obtain the first enhanced frame P ₀ ^′ .

For an image data or video data, after the encoding spectrum adjustment, the image or video may become blurred, and the amount of data information in the image or video data becomes smaller, so that the difference between adjacent pixels becomes smaller. When the image data or video data is encoded using the standard H.264/H.265, the details of some images and videos may be lost to a certain extent. That is to say, for those boundaries where the difference between adjacent pixels is small, after the encoding process, the difference between them may become smaller or even disappear, resulting in loss of details in image data or video data. . Therefore, in order to avoid the loss of borders with small differences between adjacent pixels in the encoding and decoding process, it is necessary to perform the border adjustment on those borders with small differences between adjacent pixels to perform border enhancement, so that the border is enhanced. After encoding and decoding, the details can still be preserved. For those boundaries with large differences between adjacent pixels, even after the encoding spectrum adjustment process, the remaining boundaries are still large enough. It will not disappear after encoding and decoding processing. Therefore, border adjustment may not be performed for those borders where the difference between adjacent pixels is large.

FIG. 5A shows a flowchart of acquiring a first enhanced frame according to an embodiment of the present specification. As shown in Figure 4A and Figure 5A, step SA242 may include:

SA242-2: Adjust the initial frame P ₀ through the first adjustment function H _L1 (f) to obtain the first frame.

For the convenience of description, we define the data in the first frame as P _L1 . The first adjustment function H _L1 (f) can be a low-pass frequency wave filter in the frequency domain, so that the amplitude of the initial frame P ₀ is smoothly reduced in the frequency domain, so that the initial frame P ₀ in the frequency domain is The components in the low frequency region are preserved and the components in the middle to high frequency region are attenuated, resulting in the first frame P _L1 . The first frame P _L1 is a blurred image. The first adjustment function H _L1 (f) may be a low-pass filter with smooth transition in any form, which is not limited in this specification.

SA242-4: Calculate the difference between the initial frame P ₀ and the first frame P _L1 to obtain the first boundary.

For the convenience of description, we define the data in the first boundary as P _E1 . The first boundary _PE1 includes boundary information in the initial frame _P0 . The first boundary P _E1 can be expressed as the following formula:

P _E1 =P ₀ -P _L1 =P ₀ -P ₀ *H _L1 (f) Formula (1)

The intermediate frequency to high frequency components in the data spectrum of each frame are mainly concentrated in the area where the data changes sharply in this frame of data, that is, the boundary data of the data. For example, for a frame of image, the intermediate frequency to high frequency data are mainly concentrated on the boundary of the object in the image, that is, the boundary data of this frame of image. The first adjustment function H _L1 (f) smoothly reduces the amplitude of the initial frame P ₀ in the frequency domain to attenuate the components in the mid-frequency to high-frequency region. Therefore, the first frame P _L1 can be understood as data from which the boundary information in the initial frame P ₀ is removed. Next, the difference between the initial frame P ₀ and the first frame P _L1 can be calculated to obtain the boundary of the initial frame P ₀ , that is, the first boundary P _E1 .

In some embodiments, step SA242 may further include:

_SA242-6 : Enhance the first boundary PE1 by the first coefficient a.

Wherein, the first coefficient a is any number greater than 1. In some embodiments, the first boundary P _E1 may be data obtained by calculating the difference between the initial frame P ₀ and the first frame P _L1 . In other embodiments, the first boundary _PE1 may be a boundary enhanced by the first coefficient a. At this time, the first boundary P _E1 can also be expressed as the following formula:

P _E1 =a*(P ₀ -P _L1 )=a*(P ₀ -P ₀ *H _L1 (f)) Formula (2)

As described above, when the boundary adjustment is performed on the boundary of the initial frame for adjustment, the boundary adjustment is only performed on the boundary with a smaller difference between adjacent pixels. In order to prevent the boundary adjustment from affecting other boundaries that do not need to be adjusted, the data compression device 200 may first perform signal amplification on the first boundary _PE1 by a first coefficient a greater than 1.

_SA242-8 : Through the first gamma algorithm, adjust the boundary of the first boundary whose PE1 boundary value is within the first preset range to obtain an enhanced boundary.

For the convenience of description, we define the data in the enhanced boundary as _PE . As mentioned above, when the boundary adjustment is performed on the boundary of the initial frame, the boundary adjustment is only performed on the boundary with a smaller difference between adjacent pixels. The first preset range may be a boundary value for which the boundary adjustment needs to be performed. The boundary value may be a value corresponding to each pixel in _PE1 in the first boundary. Specifically, the first preset range may be a range between [-R1, R1]. R1 can be the boundary threshold. For example, R1 could be 30, 40, 50, etc. In some embodiments, R1 can be any number between 5-30.

FIG. 5B shows a flowchart of performing the boundary adjustment according to an embodiment of the present specification. As shown in FIG. 5B, step SA242-8 may be: the data compression device 200 adjusts the first boundary P _E1 through the second adjustment function H _L2 (f) to obtain a second boundary P _E2 ; The gamma algorithm adjusts the boundary where the boundary value of the second boundary PE2 is within the first preset range [ _-R1 , R1] to obtain the enhanced boundary _PE .

In the boundary of the initial frame (the first boundary P _E1 ), the boundary that does not need to perform the boundary enhancement contains many components in the middle to high frequency region. Therefore, in order to prevent the boundary enhancement from affecting other boundaries that do not need to be adjusted, the data compression device 200 may first filter the first boundary _PE1 to filter the components in the mid-frequency to high-frequency region. The second adjustment function H _L2 (f) may be a low-pass filter with a DC component DC equal to 1, so that the components in the low frequency region of the first boundary P _E1 in the frequency domain are retained and the components in the intermediate frequency to high frequency region is filtered.

FIG. 6 shows a schematic diagram of a second adjustment function H _L2 (f) provided according to an embodiment of the present description. The horizontal axis is the normalized frequency f, and the vertical axis is the amplitude adjustment gain H _L2 of the second adjustment function H _L2 (f). The normalized frequency f of the horizontal axis can be divided into a low frequency region, a middle and low frequency region, a middle frequency region, a middle and high frequency region and a high frequency region. As shown in Fig. 6, the normalized frequency maximum value of the horizontal axis is 0.5. As mentioned above, the high frequency region may include frequencies between (d, 0.5] in the normalized frequency domain. Where d is the lower frequency limit of the high frequency region. For example, d may be the normalized frequency range. The intermediate frequency region may include frequencies between (b, c], where b is the frequency lower limit of the intermediate frequency region, and c is the frequency upper limit of the intermediate frequency region. For example, the frequency lower limit b of the intermediate frequency region may be 0.15, 0.16, 0.17, 0.18, Any frequency among 0.19, 0.2, 0.21, 0.22, 0.23, 0.24, 0.25, 0.26, 0.27 and 0.28; the upper frequency limit c of the intermediate frequency region may be 0.35, 0.34, 0.33 in the normalized frequency domain , 0.32 and 0.31 in any frequency. The low frequency region may include frequencies between [0, a] in the normalized frequency domain. Where a is the upper frequency limit of the low frequency region. The frequency of the low frequency region The upper limit a may be any one of 0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.10, 0.12, 0.13, 0.14 and 0.15 in the normalized frequency domain. When the low frequency When the intermediate frequency area is not connected with the mid-frequency area, the frequency area between the two is called the mid-low frequency area. When the intermediate frequency area is not connected with the high-frequency area, the frequency area between the two is called the mid-high frequency area. frequency area.

The second adjustment function H _L2 (f) can filter the components in the mid to high frequency region. The stop band interval in the second adjustment function H _L2 (f) may be any interval between the frequencies of 0.25 and 0.50. For example, the stop band interval in the second adjustment function H _L2 (f) can be specified by any two of the values 0.25, 0.27, 0.29, 0.31, 0.33, 0.35, 0.37, 0.39, 0.41, 0.43, 0.45 and 0.50. within the interval. The passband interval in the second adjustment function H _L2 (f) can be any interval between the frequencies of 0 and 0.35. For example, the passband interval in the second adjustment function H _L2 (f) may be 0, 0.02, 0.04, 0.06, 0.08, 0.10, 0.12, 0.14, 0.15, 0.17, 0.19, 0.21, 0.23, 0.25, 0.27, 0.29, Within the interval specified by any two of the values 0.21, 0.23 and 0.35.

FIG. 7 shows a schematic diagram of a curve of a gamma algorithm provided according to an embodiment of the present specification. The gamma algorithm is a nonlinear image and video brightness adjustment method, and its normalized function curve is shown in FIG. 7 . The gamma algorithm is an algorithm for dynamically adjusting pixel values of an image. As shown in FIG. 7 , the horizontal axis is the value before being adjusted by the gamma algorithm, and the vertical axis is the value after being adjusted by the gamma algorithm. Among them, curve 7 represents a curve with gamma value γ>1. Curve 8 represents a curve with gamma value γ=1. Curve 9 represents a curve for γ<1. When γ>1, after the image is corrected by the gamma algorithm, the absolute value of the overall gray value becomes smaller. When γ<1, after the image is corrected by the gamma algorithm, the absolute value of the overall gray value becomes larger. The gamma algorithm shown in FIG. 7 is an extended gamma algorithm. That is, taking the origin (0, 0) as the symmetry point, the gamma algorithm is centrally symmetric and extended to the third quadrant.

In step SA242-8, the data compression apparatus 200 may perform boundary correction on the boundary of the initial frame through the first gamma algorithm. The data compression apparatus 200 may perform boundary adjustment on the pixels in the second boundary _PE2 one by one. Specifically, the data compression device 200 may compare the boundary value corresponding to each pixel in the second boundary _PE2 with the boundary threshold R1; when the boundary value is within [-R1, R1], the first boundary value is used The gamma algorithm adjusts the boundary value; when the boundary value is outside [-R1, R1], no adjustment is made.

In some embodiments, the data compression apparatus 200 may perform boundary adjustment on the second boundary _PE2 using the first gamma algorithm for boundary enhancement. At this time, the first gamma algorithm may be a gamma algorithm with a gamma value γ<1. The data compression apparatus 200 may perform boundary enhancement on the pixels in the second boundary _PE2 one by one. Specifically, the data compression device 200 may compare the boundary value corresponding to each pixel in the second boundary _PE2 with the boundary threshold R1; when the boundary value is within [-R1, R1], the first boundary value is used The gamma algorithm enhances the boundary value to increase the absolute value of the boundary value; when the boundary value is outside [-R1, R1], no enhancement is performed. The data compression apparatus 200 may perform the boundary adjustment using a table look-up method. Table 1 may be stored in the data compression apparatus 200 . Table 1 stores the correspondence between the input boundary value and the output boundary value in the first gamma algorithm, that is, the correspondence between the boundary value before adjustment and the boundary value after adjustment. Taking R1=5 as an example, Table 1 can be expressed as:

In some embodiments, the data compression device 200 may further perform boundary adjustment on the second boundary _PE2 through the first gamma algorithm to reduce the boundary, so as to perform noise reduction. At this time, the first gamma algorithm may be a gamma algorithm with a gamma value γ>1. Specifically, the data compression device 200 may use a gamma algorithm with a gamma value γ>1 to weaken the boundary within the first preset range [ _-R1 , R1] in the second boundary PE2 to reduce The absolute value of the boundary value, thereby eliminating noise in the second boundary _PE2 and increasing the clarity of the initial data. The noise in the image mostly exists in the boundary where the boundary value is small. By weakening the boundary within the first preset range [-R1, R1], noise can be effectively eliminated.

In some embodiments, the data compression apparatus 200 may perform boundary adjustment on the second boundary _PE2 through the first gamma algorithm to simultaneously perform the boundary enhancement and the boundary reduction on the second boundary _PE2 , to obtain an enhanced and denoised image, which can be used for preprocessing before image compression to obtain better image quality. At this time, the first gamma algorithm may be a gamma with a gamma value γ<1 in [-R1, -R2] and [R2, R1], and a gamma value γ>1 in [-R2, R2] Horse Algorithm. Wherein, R2<R1. The data compression apparatus 200 may perform boundary adjustment on the pixels in the second boundary _PE2 one by one. Specifically, the data compression device 200 may compare the boundary value corresponding to each pixel in the second boundary _PE2 with the boundary thresholds R1 and R2; when the boundary value is between [-R1, -R2] or [R2, R1] When the boundary value is within, use the gamma algorithm with gamma value γ<1 to enhance the boundary value to make the absolute value of the boundary value larger; when the boundary value is within [-R2, R2], use The gamma algorithm with gamma value γ>1 weakens the boundary value to reduce the absolute value of the boundary value, thereby eliminating noise in the second boundary PE2; when the boundary value is between [ _-R1 , R1] outside, no adjustment is made. The data compression apparatus 200 may perform the boundary enhancement and the boundary weakening on the second boundary _PE2 according to Table 2. Taking R1=5 and R2=1 as an example, Table 2 can be expressed as:

As shown in Table 2, when the boundary value is within [-1, 1], the boundary weakening is performed for the gamma value γ>1 for noise reduction. The boundary enhancement is performed when the boundary value is within [-5, -1] and [1, 5] with a gamma value γ<1.

FIG. 5C shows another flowchart of performing boundary adjustment according to an embodiment of the present specification. As shown in FIG. 5C, step _SA242-8 may also be: the data compression device 200 directly determines that the boundary value in the first boundary PE1 is in the first preset range [- The boundary within R3, R3] is adjusted to obtain the enhanced boundary _PE . Wherein R3 can be different from R1 or can be the same as R1.

Step SA242 may also include:

_SA242-9 : Superimpose the first frame P _L1 and the enhancement boundary PE to obtain the first enhancement frame P′ ₀ .

Step SA240 may also include:

SA244: Perform the encoding spectrum adjustment and the encoding on the first enhanced frame P' ₀ .

In step SA244, the data compression device 200 may first perform the encoding spectrum adjustment on the first enhanced frame P' ₀ , so that the amplitude of the first enhanced frame P' ₀ in the frequency domain is reduced smoothly, so that the The boundary information of the first enhanced frame P' ₀ is blurred, and a coded spectrum adjustment frame is obtained to reduce the amount of information in the first enhanced frame P' ₀ , thereby reducing the occupation of the first enhanced frame P' ₀ after compression space resources. Then, encode the encoded spectrum adjustment frame, that is, predict and calculate the residual, and predict the encoded spectrum adjustment frame to obtain the predicted frame and the predicted data PI of the encoded spectrum adjustment frame; The predicted frame of the adjustment frame is subtracted from the initial frame of the encoded spectrum adjustment frame to obtain the residual data R of the encoded spectrum adjustment frame, and the residual data R and the predicted data PI are input into the code stream generation module to synthesize, to obtain the compressed frame. The data processing method P200 can improve the coding efficiency of the coded spectrum adjustment frame, further reduce the amount of data in the compressed frame, improve the coding efficiency, and improve the compression ratio. Since the object of the encoding spectrum adjustment is the first enhanced frame P′ ₀ , the under-compressed frame is the first enhanced frame P′ ₀ . Taking video data as an example, step SA244 may include:

SA244-2: Determine the frame type of the initial frame.

As mentioned above, when encoding video data using the H.264 or H.265 standard, frames are often compressed into different frame types according to frame images. Therefore, before performing the encoding spectrum adjustment on the compressed frame (the first enhanced frame P′ ₀ ), the data compression device 200 needs to determine the frame type of the initial frame, and the encoding volume selected for different frame types The nuclei are also different.

For the video frame sequence, the specific frame types may include an intra-frame prediction frame (Intra Picture, referred to as an I frame), a forward prediction frame (Predictive Frame, referred to as a P frame), and a bi-directional predictive frame (Bi-directional Predictive Frame, referred to as frame) B frame). For a frame sequence with only one frame, it is usually processed as an intra-frame predicted frame (I-frame). An I-frame is a fully intra-compressed coded frame. During decoding, only the data of the I frame can be used to reconstruct the complete data without referring to other pictures, which can be used as the reference frame of several subsequent frames. P-frames are coded frames that compress the amount of transmitted data by substantially reducing temporal redundancy with previously coded frames in the image sequence. The P frame is predicted from the P frame or I frame before it, and it compresses the frame according to the difference between the frame and the adjacent previous frame or frames. The combined compression method of P frame and I frame can achieve higher compression without obvious compression traces. It only refers to the preceding I-frame or P-frame that is close to it. The B frame compresses the current frame according to the difference between the adjacent previous frames, the current frame and the subsequent frames, that is, only the difference between the current frame and the previous and subsequent frames is recorded. Generally, the compression efficiency of the I frame is the lowest, the P frame is higher, and the B frame is the highest. During the encoding process of video data, some video frames will be compressed into I frames, some will be compressed into P frames, and some will be compressed into B frames. The frame type of the initial frame includes at least one or more of an I frame, a P frame, and a B frame.

SA244-4: Based on the frame type of the initial frame, select a convolution kernel from the encoding convolution kernel group as the encoding convolution kernel, and perform convolution on the under-compression frame to obtain an encoded spectrum adjustment frame.

Specifically, step SA244-4 may be to perform the encoding spectrum adjustment on the under compression frame (the first enhanced frame P' ₀ ) to obtain the encoded spectrum adjustment frame. Wherein, the encoding spectrum adjustment includes using an encoding convolution check to perform convolution on the under-compressed frame, so as to smoothly reduce the amplitude of the under-compressed frame in the intermediate frequency region in the frequency domain.

Performing spectral adjustment on the under-compressed frame may be expressed as multiplying the under-compressed frame by a transfer function H ₁ (f) (ie, an encoding spectrum adjustment function) in the frequency domain or performing corresponding convolution calculation in the time domain. If the compressed frame is digitized data, the convolution operation may be to select an encoding convolution kernel corresponding to the encoding spectrum adjustment function H ₁ (f) to perform the convolution operation. For the convenience of description, this specification will take convolution in the time domain as an example to describe the spectrum adjustment, but those skilled in the art should understand that the spectrum adjustment is performed by multiplying the encoded spectrum adjustment function H ₁ (f) in the frequency domain. It is also the scope of protection in this manual.

As mentioned above, performing the encoding spectrum adjustment on the under-compressed frame may be represented by convolution of the under-compressed frame in the time domain. The storage medium of the data compression device 200 may store a plurality of coded spectrum adjusters, that is, the code spectrum adjuster group. Each coded spectral conditioner includes a set of coded convolution kernels. That is, the storage medium of the data compression device 200 may include the encoding convolution kernel group, and the encoding convolution kernel group may include at least one convolution kernel. When the data compression device 200 performs convolution on the compressed frame, it may select a convolution kernel from the encoding convolution kernel group as the encoding convolution based on the frame type of the compressed frame corresponding to the initial frame. Kernel, convolution on the pressed frame. When the compressed frame corresponding to the initial frame is an I frame or a P frame, the data compression device 200 convolving the I frame or the P frame includes selecting a convolution kernel from the encoding convolution kernel group as the The encoding convolution kernel is used to perform convolution on the I frame or the P frame. Any one of the convolution kernels in the convolution kernel group can reduce the amplitude of the I frame or the P frame in the frequency domain, and reduce the amplitude in the intermediate frequency region smoothly. The data compression device 200 may also select a convolution kernel with the best compression effect from the encoding convolution kernel group as the encoding convolution kernel according to the encoding quality requirement for the initial frame. When the compressed frame corresponding to the initial frame (that is, the first enhanced frame in this embodiment) is a B frame, the encoding convolution kernel of the compressed frame is the same as the one closest to the compressed frame. The coding convolution kernels corresponding to the reference frames are the same, or the coding convolution kernels in the frame compression are the same as the coding convolution kernels corresponding to the reference frames with the largest attenuation degree among the closest reference frames in the two adjacent directions. , or the coding convolution kernel of the frame compression takes the average value of the coding convolution kernels corresponding to the nearest reference frames in two adjacent directions. When the distance between the B frame and two adjacent reference frames is the same, the coding convolution kernel in the frame compression may take the coding convolution kernel corresponding to any one of the adjacent reference frames in the two directions, for example The current B frame selects the convolution kernel corresponding to the forward adjacent reference frame. In this way, the effect of reducing the amplitude of the compressed frame is better, and the effect of adjusting the encoding spectrum is better, so that the compression ratio of the video data is higher.

FIG. 8A shows a graph of an encoding spectrum adjustment function H ₁ (f) provided according to an embodiment of the present specification. As shown in FIG. 8A , the horizontal axis is the normalized frequency f, and the vertical axis is the amplitude adjustment gain H ₁ of the encoding spectrum adjustment function H ₁ (f). Curve 1 and curve 2 in FIG. 8A represent different encoding spectral adjustment functions H ₁ (f) corresponding to different encoding convolution kernels. Taking video data as an example, since the human eye is more sensitive to low-frequency to medium-frequency data than high-frequency data, when performing the encoding spectrum adjustment on video data, the low-frequency to medium-frequency information contained in the initial frame should be preserved as much as possible. Without loss, keep the amplitude gain in the mid-frequency and low-frequency regions relatively stable, and make the information in the low-frequency to mid-frequency region as relatively stable and complete as possible, so that the information in the low-frequency to mid-frequency region can be better recovered during decompression. Therefore, the encoding spectrum adjustment function H ₁ (f) used in the encoding spectrum adjustment may be greater than zero for the amplitude adjustment gain H ₁ at any frequency f in the low frequency to the intermediate frequency region of the frame compression in the frequency domain, The amplitudes of all frequencies in the low-frequency to medium-frequency region processed by the encoding spectrum adjustment function H ₁ (f) are also greater than zero, and no data of any frequency is lost in the low-frequency to medium-frequency region. Therefore, when the compressed data is decompressed, the data in all frequency ranges from the low frequency to the intermediate frequency region can be recovered. Otherwise, if there is a zero point in the low frequency to intermediate frequency region in the encoding spectrum adjustment function H ₁ (f), the data of the frequency part corresponding to the zero point may be lost, and the decoding end will not be able to recover the lost data during decompression, so the original data cannot be recovered. . As mentioned above, we define the data obtained after the first enhanced frame P′ ₀ is processed by the encoded spectrum adjustment function H ₁ (f) as P ₁ , therefore, the data of the encoded spectrum adjustment frame is defined as is P ₁ . Since P' ₀ only enhances P ₀ in a small range, the region outside the first preset range is not enhanced, and the region P' ₀ outside the first preset range is consistent with P ₀ . Therefore, the relationship between P ₀ and P ₁ can be expressed as the following formula:

P ₁ =H ₁ (f)·P′ ₀ ≈H ₁ (f)·P ₀ Formula (3)

Since the human eye is relatively insensitive to high-frequency data, when the encoding spectrum adjustment is performed on the video data, the amplitude of the high-frequency part can be attenuated to a greater degree, and the amplitude of the high-frequency region can be reduced to a greater degree. value. In this way, the data information contained in the first enhanced frame can be reduced, and the compression ratio and coding efficiency can be improved.

Therefore, the encoding spectrum adjustment function H ₁ (f) used for the encoding spectrum adjustment can smoothly reduce the amplitude of the frame under compression in the frequency domain. In some embodiments, the encoding spectrum adjustment function H ₁ (f) used by the encoding spectrum adjustment can smoothly reduce the amplitude of the high frequency region of the frame in its frequency domain. The steady decrease of the amplitude value may be that the amplitude value is attenuated by the first amplitude value adjustment gain value in the high frequency region, or it may be a certain error of the amplitude value near the first amplitude value adjustment gain value. attenuation within the range. For example, the first amplitude adjustment gain may be any value between 0 and 1. For example, the first amplitude adjustment gain may be 0, 0.04, 0.08, 0.12, 0.16, 0.20, 0.24, 0.28, 0.32, 0.36, 0.40, 0.44, 0.48, 0.52, 0.56, 0.60, 0.64, 0.68, 0.72, Within the interval specified by any two of the values 0.76, 0.80, 0.84, 0.88, 0.92, 0.96 and 1. The error range can be 0, ±1%, ±2%, ±3%, ±4%, ±5%, ±6%, ±7%, ±8%, ±9%, ±10%, ±11 %, ±12%, ±13%, ±14%, ±15%, ±16%, ±17%, ±18%, ±19%, ±20%, ±21%, ±22%, ±23%, Within the interval specified by any two values of ±24%, ±25%, ±26%, ±27%, ±28%, ±29%, ±30%, etc. As shown in FIG. 8A , the first amplitude adjustment gain of the encoding spectrum adjustment in the high frequency region (approximately the interval of 0.4 to 0.5) is about 0.2.

In some embodiments, the encoding spectrum adjustment function H ₁ (f) used in the encoding spectrum adjustment can smoothly reduce the amplitude in the intermediate frequency region of the frame in the frequency domain. Wherein, the amplitude adjustment gain of the encoding spectrum adjustment to the intermediate frequency region in the frame compression is the second amplitude adjustment gain. In some embodiments, the value of the second amplitude adjustment gain may be greater than the first amplitude adjustment gain, as shown in FIG. 8A . When the coded spectrum is adjusted to be frequency attenuated (that is, when the coded spectrum adjuster is the frequency attenuator), both the first amplitude adjustment gain and the second amplitude adjustment gain are less than 1. That is to say, the magnitude of the reduction of the amplitude of the intermediate frequency region of the compressed frame by the encoding spectrum adjustment may be lower than the magnitude of the reduction of the amplitude of the high frequency region.

In addition, the encoding spectrum adjustment function H ₁ (f) can also smoothly reduce the amplitude in the low frequency region of the frame in the frequency domain. Wherein, the amplitude adjustment gain of the encoding spectrum adjustment to the low frequency region of the frame compression is a third amplitude adjustment gain. When the coded spectrum is adjusted to be frequency attenuated (that is, when the coded spectrum adjuster is the frequency attenuator), both the third amplitude adjustment gain and the second amplitude adjustment gain are less than 1. The value of the third amplitude adjustment gain may be greater than or equal to the second amplitude adjustment gain. That is to say, the magnitude of reduction of the amplitude of the low-frequency region in the compressed frame by the encoding spectrum adjustment may be lower than or equal to the magnitude of the reduction of the amplitude of the intermediate-frequency region.

Further, in order to save the amount of calculation required in the implementation process and avoid the occurrence of ringing effects, the encoding spectrum adjustment function H ₁ (f) should make the amplitude of the first enhanced frame P′ ₀ in the frequency domain smoothly. transition. As mentioned above, when performing spectrum adjustment processing on an image, if the selected spectrum adjustment function has a region with dramatic changes in value, a convolution kernel or a combination of convolution kernels with a higher order is required in the implementation process. This means adding unnecessary computation. At the same time, high-order convolution kernels are more likely to produce strong color oscillations in the output image where the grayscale or color changes sharply, which is called the ringing effect. The ringing effect mostly occurs at the image boundary. The sharp change of the amplitude adjustment gain can be avoided by making the coded spectral adjustment function H ₁ (f) smoothly transition the amplitude adjustment gain of the first enhancement frame P′ ₀ in the frequency domain. For example, when the high frequency region is not connected to the intermediate frequency region, the encoding spectrum adjustment function H ₁ (f) can adjust the amplitude of the middle and high frequency region in the frame in the frequency domain, The variation of the amplitude adjustment gain in the middle and high frequency region is made smooth and continuous. When the intermediate frequency region is not connected to the low frequency region, the encoding spectrum adjustment function H ₁ (f) can adjust the amplitude of the middle and low frequency regions in the frame in the frequency domain, so that the The amplitude adjustment gain changes continuously in the mid-low frequency region.

The coding spectrum adjustment function H ₁ (f) can also maintain the DC part, that is, the amplitude adjustment gain in the part where the frequency is 0 is 1, so as to ensure that the basic information in the first enhanced frame P′ ₀ can be preserved. When the data is decompressed, the average information can be obtained to restore the original original data. Therefore, the amplitude reduction range of the coding spectrum adjustment function H ₁ (f) used in the coding spectrum adjustment for the low frequency region is lower than that of the intermediate frequency region. However, when the amplitude gain of the DC part (ie, the part with a frequency of 0) is not 1, the original data can also be recovered by designing an appropriate decoding spectrum adjustment function H ₂ (f). Specifically, the specific relationship between H ₁ (f) and H ₂ (f) will be introduced in detail in the following description.

In the graph of the encoding spectrum adjustment function H ₁ (f) shown in FIG. 8A , the frequencies between (0, 0.1] belong to the low frequency; the frequencies between (0.1, 0.15] belong to the middle and low frequencies; (0.15, 0.33] The frequency between (0.33, 0.4] belongs to the middle and high frequency; the frequency between (0.4, 0.5] belongs to the high frequency. The third amplitude adjustment gain in the low frequency region is greater than that in the intermediate frequency region. The second amplitude adjustment gain; the second amplitude adjustment gain in the intermediate frequency region is greater than the first amplitude adjustment gain in the high frequency region. At the same time, the second amplitude adjustment gain in the intermediate frequency region is relatively stable, curve 1 At around 0.5, curve 2 is around 0.6; the first amplitude adjustment gain H1 in the high frequency region is also relatively stable, curve ₁ is slightly lower than 0.2, and curve 2 is slightly higher _than 0.2. The encoding spectrum adjustment function H1 The curve of (f) is a curve of smooth transition.In engineering implementation, on the basis of reducing the amplitude, it is possible to allow the curve of the encoding spectrum adjustment function H ₁ (f) to have a small range of fluctuations, and the fluctuation does not affect The effect of compression.For other forms of data other than video data, the parameters of the encoding spectrum adjustment function H ₁ (f) can be set according to the sensitivity of the receiver to the data. For different forms of data, the receiver is sensitive to the frequency sensitivities vary.

FIG. 8B shows a graph of an encoding spectrum adjustment function H ₁ (f) provided according to an embodiment of the present specification. Curve 3 and curve 4 in FIG. 8B represent different encoding spectral adjustment functions H ₁ (f) corresponding to different encoding convolution kernels. As far as video data is concerned, in some special application scenarios, it is necessary to properly retain more high-frequency components, such as reconnaissance scenes. Therefore, in some embodiments, the encoding spectrum adjustment function H ₁ (f) curve can make the first amplitude adjustment gain greater than the second amplitude adjustment gain (curve 3), or equal to the second amplitude adjustment gain (curve 4 ).

As far as video data is concerned, in some application scenarios that do not require high image quality, high-frequency components can be completely filtered out. Therefore, the encoding spectrum adjustment function H ₁ (f) used in the encoding spectrum adjustment The amplitude adjustment gain H1 _of any frequency from the low frequency to the intermediate frequency region of the frame in the frequency domain is greater _than zero, and the amplitude adjustment gain H1 for the high frequency region can be equal to 0 (not shown in FIG. 8A and FIG. 8B ). ).

或者编码频谱调节函数乘积组合

或者线性组合和乘积组合的组合都属于本说明书保护的范围。其中，i≥1，

代表n个函数的线性组合，H_1i(f)代表第i个函数，k_i代表第i个函数对应的权重。j≥1，

代表n个函数的乘积组合，k_j代表第j个函数对应的权重，H_1j(f)可以是任意函数。It should be noted that the curves shown in FIG. 8A and FIG. 8B are only illustrated by taking video data as an example, and those skilled in the art should understand that the curve of the encoding spectrum adjustment function H ₁ (f) is not limited to FIG. 8A and the form shown in FIG. 8B , all the coding spectrum adjustment functions H ₁ (f) and the linear combination of the coding spectrum adjustment functions that can make the amplitude of the intermediate frequency region of the first enhanced frame in the frequency domain smoothly decrease

Or code spectrum adjustment function product combination

Or the combination of linear combination and product combination all belong to the protection scope of this specification. Among them, i≥1,

Represents the linear combination of n functions, H _1i (f) represents the _ith function, and ki represents the weight corresponding to the ith function. j≥1,

Represents the product combination of n functions, k _j represents the weight corresponding to the jth function, and H _1j (f) can be any function.

Table 3 shows a parameter table of a coding convolution kernel provided according to an embodiment of the present specification. Table 3 exemplarily lists the parameters of an encoding convolution kernel, wherein each row in Table 3 represents an encoding convolution kernel. For an 8-bit video image, it is necessary to ensure that the gray value of the pixel points in the encoded spectrum adjustment frame obtained after encoding convolution is within 0 to 255. Therefore, in this embodiment, the convolution result needs to be Divide by 256. The encoding convolution kernel is obtained by Fourier transform based on the encoding spectral adjustment function H ₁ (f). Table 3 is only an exemplary illustration. Those skilled in the art should know that the coding convolution kernel is not limited to the parameters shown in Table 3, all the parameters that can make the amplitude of the intermediate frequency region of the frame in the frequency domain stabilize The coding convolution kernels with low level reduction all belong to the protection scope of this specification.

It should be noted that, in order to avoid the ringing effect, the encoding spectrum adjustment function H ₁ (f) is a smooth transition curve, so as to avoid the sharp change of the amplitude adjustment gain in the curve. As mentioned above, the ringing effect refers to that in image processing, when a spectrum adjustment process is performed on an image, if the selected spectrum adjustment function has a rapid change, the image will cause "ringing". The so-called "ringing" refers to the vibration generated by the sharp change of the grayscale of the output image, just like the air vibration generated by a clock being struck. The ringing effect mostly occurs at the image boundary.

The ratio of the absolute value of the sum of the negative coefficients to the sum of the non-negative coefficients in the coding convolution kernel corresponding to the coding spectrum adjustment function H ₁ (f) should be less than 0.1. For example, in some embodiments, the coefficients of the convolution kernel in the encoding convolution kernel may all be non-negative numbers. Taking video data as an example, when there are many negative coefficients in the encoding convolution kernel, the pixel values at the image boundaries are very different. A large pixel value multiplied by a negative coefficient will make the final convolution result. The result is smaller, which is reflected in the image as darker pixels. If the convolution result has a negative number and the absolute value of the negative number is large, when the unsigned integer calculation is used to calculate the convolution result, the unsigned integer calculation result may be reversed, and the value is the unsigned complement value of the negative number, then It will cause the convolution result to become larger, which is reflected in the image as brighter pixels. Therefore, when designing the encoding convolution kernel, the coefficients of the encoding convolution kernel may be all non-negative numbers, or the absolute value of the sum of the negative coefficients in the encoding convolution kernel may be the same as the sum of the non-negative coefficients. The ratio of the sum should be less than 0.1, that is, a small number of negative coefficients with small absolute values are allowed to appear in the encoding convolution kernel.

When the data compression device 200 uses the encoding convolution check to perform convolution on the pressed frame, the data compression device 200 may perform the convolution on the pressed frame (initial frame) in at least one of a vertical direction, a horizontal direction, and an oblique direction. product.

It should be pointed out that when performing the above-mentioned convolution in frame compression, the data processing unit processed may be a frame of data, or may be a part of a frame of data. Taking video data as an example, the unit can be a frame or field image, or a part of a frame/field image. For example, in video coding, the image is further divided into slices, tiles, and coding units ( coding unit, CU), macroblock, or block. The convolution object includes but is not limited to a part of the image segmentation unit described by the above terms. In different processing units, the same encoding convolution kernel can be selected, or different encoding convolution kernels can be selected.

Step SA244 may also include:

SA244-6: Perform the encoding (prediction and residual error calculation) on the encoded spectrum adjustment frame to obtain the predicted data PI and the residual data R.

SA244-8: Input the prediction data PI and the residual data R into the code stream generation module for synthesis to obtain the compressed frame.

After the data compression device 200 performs the encoding spectrum adjustment on the first enhanced frame, the encoded spectrum adjustment frame is obtained, and the frequency components from low frequency to high frequency in the encoded spectrum adjustment frame are smaller than the first enhanced frame low to high frequency components in the middle. Therefore, the data compression device 200 can improve the encoding efficiency of the encoded spectrum adjustment frame by performing encoding and code stream generation calculation on the compressed frame (first enhanced frame) after performing the encoding spectrum adjustment, thereby improving the The compression ratio of the initial frame improves the transmission efficiency of the initial data; at the same time, the boundary enhancement can avoid loss of details.

The data processing method PB200 shown in FIG. 4B corresponds to the flowchart shown in FIG. 3B . The method PB200 shown in FIG. 4B may include:

SB220: Select the initial frame in the initial data. It is the same as step SA220, and will not be repeated here.

SB240: Perform the data compression on the initial frame to obtain a compressed frame. Step SB240 may include:

SB242: Perform the boundary adjustment on the initial frame P ₀ to obtain the first enhanced frame P′ ₀ . It is the same as step SA242, and will not be repeated here.

SB244: Perform the encoding spectrum adjustment and the encoding on the first enhanced frame P' ₀ . Step SB244 may include:

SB244-2: Determine the frame type of the initial frame. It is the same as step SA244-2, and will not be repeated here.

SB244-4: Predict the first enhanced frame P′ ₀ to obtain a predicted frame and predicted data PI.

SB244-6: Based on the frame type of the initial frame, select a convolution kernel from the encoding convolution kernel group as the encoding convolution kernel, and perform convolution on the under-compression frame to obtain an encoded spectrum adjustment frame. Wherein, the under-compressed frame includes the first enhanced frame P′ ₀ and the predicted frame. Step SB244-6 may be equivalent to performing the encoding spectrum adjustment on the first enhanced frame P' ₀ and the prediction frame using the encoding spectrum adjustment function H ₁ (f). The encoded spectrum adjustment frame includes a first encoded spectrum adjustment frame obtained by performing the encoding spectrum adjustment on the first enhanced frame P′ ₀ and a second encoded spectrum adjustment frame obtained by performing the encoding spectrum adjustment on the predicted frame. frame.

SB244-8: Obtain a residual difference between the first coded spectrum adjustment frame and the second coded spectrum adjustment frame to obtain the residual data R.

SB244-9: Input the prediction data PI and the residual data R into the code stream generation module for synthesis to obtain the compressed frame.

The data processing method PC200 shown in FIG. 4C corresponds to the flowchart shown in FIG. 3C . The method PC200 shown in FIG. 4C may include:

SC220: Select the initial frame in the initial data. It is the same as step SA220, and will not be repeated here.

SC240: Perform the data compression on the initial frame to obtain a compressed frame. Step SC240 may include:

SC242: Perform the boundary adjustment on the initial frame P ₀ to obtain the first enhanced frame P′ ₀ . It is the same as step SA242, and will not be repeated here.

SC244: Perform the encoding spectrum adjustment and the encoding on the first enhanced frame P' ₀ . Step SC244 may include:

SC244-2: Determine the frame type of the initial frame. It is the same as step SA244-2, and will not be repeated here.

SC244-4: First perform the encoding on the first enhanced frame P′ ₀ , that is, predict and calculate the residual to obtain the predicted data PI and the residual data R ₁ .

SC244-6: Based on the frame type of the initial frame, select a convolution kernel from the encoding convolution kernel group as the encoding convolution kernel, perform convolution on the compressed frame, and obtain the residual data R . Wherein, the in-pressing frame includes the residual data R ₁ . Step SC244-6 may be equivalent to performing the encoding spectrum adjustment on the residual data R ₁ by using the encoding spectrum adjustment function H ₁ (f).

SC244-8: Input the prediction data PI and the residual data R into the code stream generation module for synthesis to obtain the compressed frame.

The data processing method PD200 shown in FIG. 4D corresponds to the flowchart shown in FIG. 3D . The method PD200 shown in FIG. 4D may include:

SD220: Select the initial frame in the initial data. It is the same as step SA220, and will not be repeated here.

SD240: Perform the data compression on the initial frame to obtain a compressed frame. Wherein, step SD240 may be to perform the encoding spectrum adjustment on the initial frame first, and then perform the boundary adjustment. Specifically, step SD240 may include:

SD242: Perform the encoding spectrum adjustment on the initial frame to obtain an encoded spectrum adjustment frame.

For the convenience of description, we define the data in the encoded spectrum adjustment frame obtained in step SD242 as P ₀₁ . Specifically, step SD242 may include:

SD242-2: Determine the frame type of the initial frame. It is the same as step SA244-2, and will not be repeated here.

SD242-4: Based on the frame type of the initial frame, select a convolution kernel from the encoding convolution kernel group as the encoding convolution kernel, perform convolution on the pressing frame, and obtain the encoded spectrum adjustment frame P ₀₁ . Wherein, the under-pressing frame includes the initial frame P ₀ . Step SD242-4 may be equivalent to performing the encoding spectrum adjustment on the initial frame P ₀ by using the encoding spectrum adjustment function H ₁ (f).

SD244: Perform the boundary adjustment on the encoded spectrum adjustment frame P ₀₁ to obtain a second enhanced frame.

For the convenience of description, we define the data in the second enhanced frame obtained in step SD244 as P' ₁ . Fig. 9 shows a flow chart of acquiring the second enhanced frame P' ₁ according to an embodiment of the present specification. As shown in Figure 9 and Figure 4D, step SD244 may include:

SD244-2: Calculate the difference between the initial frame P ₀ and the coded spectrum adjustment frame P ₀₁ to obtain a first boundary P _E1 .

As mentioned above, the coded spectrum adjustment can make the amplitude of the initial frame P ₀ in the frequency domain decrease smoothly, so that the boundary information of the initial frame P ₀ is blurred, and the coded spectrum adjustment frame P ₀₁ is obtained to The amount of information in the initial frame P ₀ is reduced, thereby reducing the space resources occupied by the initial frame P ₀ after being compressed. Therefore, the encoded spectrum adjustment frame P ₀₁ can be understood as data with attenuated boundary information in the initial frame P ₀ . Next, the difference between the initial frame P ₀ and the coded spectrum adjustment frame P ₀₁ can be calculated to obtain the boundary attenuation value of the initial frame P ₀ , that is, the first boundary P _E1 . The first boundary _PE1 includes boundary information of the initial frame _P0 .

In some embodiments, step SD244 may further include:

_SD244-4 : The first boundary PE1 is enhanced by the first coefficient a. This step is basically the same as step SA242-6, and will not be repeated here.

_SD244-6 : Through the first gamma algorithm, adjust the boundary of the first boundary PE1 whose boundary value is within the first preset range [-R1, R1] to obtain the enhanced boundary P _E. This step is basically the same as step SA242-8, and will not be repeated here.

_SD244-8 : Calculate the difference between the enhancement boundary PE and the first boundary _PE1 to obtain an adjustment value.

For the convenience of description, we define the data in the adjustment value as E _d . The adjustment value _Ed includes the adjustment value when the first boundary _PE1 is adjusted.

In some embodiments, step SD244-6 and step SD244-8 may be combined. Table 4 may be stored in the data compression device 200 . Table 4 stores the correspondence between the input boundary value of the first boundary _PE1 and the adjustment value _Ed . Taking R1=6 as an example, Table 4 can be expressed as:

The data compression device 200 can directly obtain the adjustment value _Ed according to Table 4.

_SD244-9 : Superimpose the coded spectrum adjustment frame P ₀₁ and the adjustment value Ed to obtain the second enhanced frame P′ ₁ .

As previously described, the encoding spectrum adjustment may blur the boundary of the encoding spectrum adjustment frame P ₀₁ . The difference between adjacent pixels in the encoded spectrum adjustment frame P ₀₁ becomes smaller. In order to avoid the loss of details, the coded spectrum adjustment frame P ₀₁ and the adjustment value _Ed are superimposed, so that the boundaries with small differences between adjacent pixels in the coded spectrum adjustment frame P ₀₁ are enhanced, thereby reducing the data in the Lost during encoding to avoid loss of details.

Step SD240 may also include:

_SD246 : Perform prediction and residual error calculation on the second enhanced frame P'1.

Perform the encoding (prediction and residual) on the second enhanced frame P' ₁ to obtain the predicted data PI and the residual data R, and combine the predicted data PI and the residual data R Input the code stream generation module for synthesis to obtain the compressed frame.

To sum up, the data processing method P200 can perform the boundary adjustment and the encoding spectrum adjustment on the initial frame at the same time, so as to improve the compression ratio of the initial frame, improve the encoding efficiency and the transmission efficiency of the initial data At the same time, it can reduce data loss and avoid loss of details.

FIG. 10 shows a flow chart of a method P300 of data processing for decompressing a compressed frame. As described above, the data decompression apparatus 300 may perform the data processing method P300. Specifically, the storage medium in the data decompression device 300 may store at least one set of instruction sets. The instruction set is configured to instruct the decompression processor in the data decompression apparatus 300 to complete the data processing method P300. When the data decompression device 300 is running, the decompression end processor can read the instruction set and execute the data processing method P300.

For the convenience of description, we will describe the method P300 of the data processing in the method shown in FIG. 3A and FIG. 3D . The method P300 may include:

S320: Obtain compressed data. The compressed data includes the compressed frame.

The compressed data may include the compressed frame obtained by performing data compression on the initial frame in the initial data by the data processing method P200. The compressed frame includes compressed prediction data PI and residual data R. As shown in FIG. 3A and FIG. 3D , step S320 may include: inputting the compressed frame into the code stream parsing module for analysis and calculation, to obtain the prediction data PI and the residual data R. As mentioned earlier, in this application, a frame is a common processing unit that composes a data sequence. In data processing, calculations are often performed in units of frames. In the data processing method P200 in which the data compression device 200 compresses data, the initial data may be compressed in units of frames. When the data decompression device 300 decompresses a compressed frame, data decompression may also be performed in units of frames.

S340: Perform data decompression on the compressed frame to obtain a decompressed frame.

The data decompression refers to performing decompression calculation on the compressed frame to obtain a decompressed frame, and restoring or substantially restoring the decompressed frame to the original data, or making the decompressed frame clearer than the original data. Taking video data as an example, when the amplitude of the decompressed frame at any frequency in the low frequency to medium frequency region is restored to the threshold value of the initial frame or above the threshold value, it is difficult for the human eye to detect that the decompressed frame is different from the decompressed frame. The difference in the initial frame. The threshold can be any value between 80%-90%. For example, the threshold may be in a closed interval defined by any two values of 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, and 90% any value of . For example, the data decompression should make the amplitude of the decompressed frame at any frequency in the low frequency to medium frequency region not less than 85%±3% of the initial frame.

The data decompression includes performing decoding spectrum adjustment on the decompressed frame, and performing further boundary correction on the data after the decoding spectrum adjustment, so as to obtain the desired decompressed frame. The deframed data, that is, a frame of data being decompressed, includes the compressed frame and any data state before the compressed frame becomes the decompressed frame during the decompression process.

Taking video data as an example, the data processing method P200 uses a combination of encoding spectrum adjustment and encoding to compress the initial frame, so as to further improve the compression ratio of video data and improve the efficiency of video transmission. In the video decompression technology, the data processing method P300 can decompress the compressed frame by a combination of decoding (that is, restoring the compressed frame according to the residual data R and the prediction data PI) and decoding spectrum adjustment to obtain the required decompression frame to recover the data in the compressed frame. The de-framing may include the compressed frame and any data state of the compressed frame in the decoding process according to the prediction data PI and the residual data R. For example, the decomposing frame may be the compressed frame, a decoded frame obtained through decoding, or a predicted frame obtained through prediction, and so on.

The decoding spectrum adjustment applied to the data decompression on the compressed frame refers to inputting the deframed input into a decoding spectrum adjuster to perform decoding spectrum adjustment. The decoding spectrum adjustment may correspond to the encoding spectrum adjustment, that is to say, there should be a preset correlation between the decoding spectrum adjustment function H ₂ (f) and the encoding spectrum adjustment function H ₁ (f). By carefully setting the correlation between the decoded spectrum adjustment function H ₂ (f) and the encoded spectrum adjustment function H ₁ (f), after the encoded spectrum adjusted compressed frame undergoes the decoded spectrum adjustment and the data processing, Without considering other calculation errors, it can completely restore or basically restore the data indicators before the encoding spectrum adjustment (such as the image clarity of the image data), and even exceed the data before the encoding adjustment in some indicators (such as the decoded image sharper than the original image). The specific correlation between the decoded spectrum adjustment function H ₂ (f) and the encoded spectrum adjustment function H ₁ (f) is related to the data processing manner of the decoded spectrum adjusted data. The data processing methods are different, and the correlation between the spectrum adjustment function H ₂ (f) and the encoded spectrum adjustment function H ₁ (f) is also different. The specific manner of the data processing and the correlation between the spectrum adjustment function H ₂ (f) and the encoded spectrum adjustment function H ₁ (f) will be specifically introduced in the following description.

Like the encoding spectrum adjustment, the decoding spectrum adjustment can also be performed by convolution in the time domain, so that the decoding spectrum adjustment function H ₂ (f) (that is, the decoding transfer function) in the frequency domain is used to adjust the deframed frame. spectrum. Therefore, there should also be a corresponding relationship between the decoding convolution kernel used in the decoding spectrum adjustment and the encoding convolution kernel used in the encoding spectrum adjustment. By selecting the decoding spectrum adjustment function H ₂ (f) and the decoding convolution kernel corresponding to the encoding spectrum adjustment function H ₁ (f) and the encoding convolution kernel, the two methods can achieve the same effect. For the convenience of description, this specification will take convolution in the time domain as an example to describe the decoding spectrum adjustment, but those skilled in the art should understand that the spectrum adjustment is performed by multiplying the decoding spectrum adjustment function H ₂ (f) in the frequency domain. The method is also the scope of protection of this specification.

As mentioned above, the encoding spectrum adjustment can attenuate the amplitude of the intermediate frequency region of the compressed frame in its frequency domain, and blur the boundary data of the compressed frame, thereby reducing the data generated by encoding. quantity. The decoding spectrum adjustment and the data processing can restore or even enhance the data after the encoding spectrum adjustment and data processing. That is, the decoding spectrum adjustment and the data processing can completely restore or substantially restore the amplitude of the sensitive frequency in the de-framing to the state before attenuation or even enhance it relative to the state before attenuation. Taking video data as an example, since human eyes are sensitive to low-frequency to intermediate-frequency information in images, the decoding spectrum adjustment and the data processing can restore or even enhance the amplitude of the low-frequency to intermediate-frequency regions in the video data. . Therefore, the amplitude of the decompressed frame in the low to medium frequency region should be at least restored or substantially restored to the amplitude of the initial frame in the low to medium frequency region. In video data, since human eyes are relatively insensitive to high-frequency data, the decoding spectrum adjustment and the data processing may not restore the amplitude of the high-frequency region, so that the amplitude of the high-frequency region remains attenuated .

As described above, the data compression operation attenuates the amplitude of the initial frame in the mid-frequency region or the mid-frequency to high-frequency region through the encoding spectrum adjustment, thereby reducing the amount of data information in the initial frame. Taking video data as an example, since the edge part of the object in the image is rich in medium frequency and high frequency information, and the medium frequency and high frequency area will carry more data, so reducing the amplitude of the medium frequency to high frequency area will visually make the At the same time, the information content in the image will be greatly reduced by blurring the data at the boundary of the pressed frame. Therefore, the data decompression can extract boundary information from the compressed frame, and perform boundary enhancement on the boundary information to restore it to the state in the original frame, or to make it different from the state in the original frame. enhanced.

There are many ways of processing the boundary enhancement. In the traditional technology, a high-pass filter or a band-pass filter is directly used to filter the compressed frame, and the components of the low-frequency region in the compressed frame are filtered out, and the compressed frame is extracted. The components in the mid-frequency to high-frequency region in the extraction of boundary information. However, many negative coefficients may appear in the coefficients of the convolution kernels corresponding to the high-pass filter and the band-pass filter. As mentioned above, when there are many negative coefficients in the convolution kernel, a strong ringing effect may appear in the image obtained by convolution with the convolution kernel. Therefore, in order to avoid the ringing effect, the data decompression described in this specification uses a smooth transition decoded spectrum adjustment function H ₂ (f) to perform spectral adjustment on the compressed frame, filtering the intermediate frequency to high frequency region in the compressed frame. Then, the difference between the compressed frame and the compressed frame adjusted by the decoded spectrum can be obtained to obtain the boundary information, and the boundary information can be adjusted using the adjustment coefficient to restore it to the original state or relative to the original state. The initial state described above has been enhanced. When using the above scheme to obtain boundary information, a decoding convolution kernel can be designed so that all coefficients are non-negative, or the ratio of the absolute value of the sum of the negative coefficients to the sum of the non-negative coefficients is less than 0.1, which can avoid ringing emergence of the effect.

Specifically, step S340 may include:

S342: Decode the compressed frame to obtain a decoded frame. In the method P300, the under-decoding frame may be the decoded frame.

The compressed frame may be obtained by encoding the spectrum adjustment frame by the data compression device 200 . The data decompression apparatus 300 may decode the compressed frame to obtain the decoded frame. That is, prediction is performed according to the prediction data PI to obtain a prediction frame, which is superimposed with the residual data R to obtain decoded data P ₂ , and the decoded data P ₂ is the data P ₂ of the decoded frame. There may be certain errors in the encoding and decoding process. Assuming that the deviation caused by the encoding and decoding process is small, the data P ₂ in the decoding frame is basically the same as the data P ₁ in the encoding spectrum adjustment frame. Therefore, P _The relationship between ₁ and P2 can be expressed as the following formula:

P ₂ ≈P ₁ Formula (4)

The data decompression apparatus 300 may perform the decoded spectrum adjustment and the boundary correction on the decoded frame. FIG. 11A shows a flowchart of the decoding spectrum adjustment and the boundary correction provided according to an embodiment of the present specification. FIG. 11A corresponds to FIG. 10 . As shown in FIG. 11A and FIG. 10 , step S340 may further include:

S344: Perform the decoded spectrum adjustment on the frame being decoded (ie, the decoded frame P ₂ ) to obtain a decoded spectrum adjustment frame.

For the convenience of description, we define the data in the decoded spectrum adjustment frame as PC _. The decoded spectral adjustment prevents ringing of the decoded frame. The decoding spectrum adjustment includes using the decoding spectrum adjustment function H ₂ (f) to perform the decoding spectrum adjustment on the de-framing, so that the amplitude of the de-framing in the frequency domain is smoothly reduced to filter the de-framing. The decoded spectrally adjusted frame is obtained from the deframed components in the mid-frequency to high-frequency region. The intermediate frequency to high frequency components in the data spectrum of each frame are mainly concentrated in the area where the data changes sharply in this frame of data, that is, the boundary data of the data. For example, for a frame of image, the intermediate frequency to high frequency data are mainly concentrated on the boundary of the object in the image, that is, the boundary data of this frame of image. Therefore, the data PC in the decoded spectrum adjustment frame can be understood as the data from which the boundary information in the decoded frame P2 is _{removed .} The data PC in the decoded spectrum adjustment frame can be expressed as the following _formula :

P _C =H ₂ (f)·P ₂ =H ₁ (f)·H ₂ (f)·P′ ₀ ≈ H ₁ (f)·H ₂ (f)·P ₀ Formula (5)

The decoding spectrum adjustment includes convolving the currently de-framed (decoded frame) with a corresponding decoding convolution kernel based on the encoding convolution kernel. In order to avoid ringing effect, the ratio of the absolute value of the sum of negative coefficients to the sum of non-negative coefficients in the decoding convolution kernel is smaller than a threshold. For example, the threshold may be any one of 0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.1, 0.2, 0.3, and 0.4, or any value in an interval defined by any two numbers. For example, the coefficients of the convolution kernel in the decoding convolution kernel may all be selected as non-negative numbers. The amplitude adjustment gain for the mid-frequency to high-frequency region in the decoding spectrum adjustment function H ₂ (f) is equal to 0, and can fluctuate within a certain error range. The error range can be 0, ±1%, ±2%, ±3%, ±4%, ±5%, ±6%, ±7%, ±8%, ±9%, ±10%, ±11 %, ±12%, ±13%, ±14%, ±15%, ±16%, ±17%, ±18%, ±19%, ±20%, ±21%, ±22%, ±23%, ±24%, ±25%, ±26%, ±27%, ±28%, ±29%, ±30%, ±31%, ±32%, ±33%, ±34%, ±35%, etc. within any two specified intervals.

By decoding the spectrum adjustment function H ₂ (f), the DC part can be maintained, that is, the amplitude adjustment gain in the part where the frequency is 0 is 1, so as to ensure that the basic information in the initial frame can be preserved. Therefore, the decoding spectrum adjustment function H ₂ (f) used in the decoding spectrum adjustment has a smooth transition to the amplitude adjustment gain of the low frequency region from the amplitude adjustment gain 1 at the position of frequency 0 to the amplitude adjustment gain of the intermediate frequency region. The value adjustment gain is close to 0.

Step S344 may include: determining the frame type of the decoded frame; based on the frame type of the decoded frame, selecting a convolution kernel from the decoding convolution kernel group as the decoding convolution kernel, and convoluting the decoded frame product.

As mentioned above, in the process of compressing the initial frame, the data compression apparatus 200 may encode the initial frame or the encoded spectrum adjustment frame into different types. Therefore, before performing the decoding spectrum adjustment on the decoded frame, the data decompression device 300 needs to determine the frame type of the decoded frame, and the decoding convolution kernels selected for different frame types are also different. The frame type of the decoded frame may include at least one of an I frame, a P frame, and a B frame. The frame type of the decoded frame may include only one frame type, or may include multiple frame types at the same time. The method for determining the frame type of the decoded frame is relatively mature, and is not the focus of the present specification, so it will not be repeated here.

As previously described, performing the decoded spectral adjustment on the decoded frame may appear as convolution of the decoded frame in the time domain. The storage medium of the data decompression device 300 may store a plurality of different decoding convolution kernels, which are called decoding convolution kernel groups. Each encoding convolution kernel corresponds to at least one decoding convolution kernel in the decoding convolution kernel group. When the data decompression device 300 performs convolution on the decoded frame, it may select a convolution kernel from the decoded convolution kernel group as the decoded convolution kernel based on the frame type of the decoded frame, and perform the decoding on the decoded convolution kernel. Frame do convolution. The operation of convolving on deframes using a decoded convolution kernel may be referred to as a decoded spectrum conditioner. When the decoded frame is an I frame or a P frame, the data decompression apparatus 300 convolving the I frame or the P frame includes selecting a convolution kernel from the decoding convolution kernel group as the decoding convolution kernel , and perform convolution on the I frame or P frame. The data decompression device 300 may also select a convolution kernel with the best decompression effect from the decoding convolution kernel group as the decoding convolution kernel according to the decoding quality requirement for the decoded frame. When the decoded frame is a B frame, the decoded convolution kernel of the decoded frame is the same as the decoded convolution kernel of the reference frame closest to the decoded frame, or the decoded convolution kernel of the decoded frame The decoding convolution kernel corresponding to the reference frame with the largest attenuation degree among the closest reference frames in the adjacent two directions is the same, or the decoding convolution kernel of the decoding frame takes the closest reference frame in the adjacent two directions. The average value of the decoded convolution kernel corresponding to the reference frame. When the distance between the decoded frame and the two nearest reference frames is the same, the decoded convolution kernel of the decoded frame takes the decoded convolution kernel of the nearest forward or backward reference frame. When the decoded frame is a B frame, the reference frame corresponding to the decoded convolution kernel selected by the decoded frame should be the reference frame corresponding to the encoded convolution kernel selected during the encoding spectrum adjustment of the compressed frame Frames are the same.

When the data decompression device 300 uses the decoding convolution check to perform convolution on the deframed frame, it may perform convolution on the deframed frame in at least one of a vertical direction, a horizontal direction, and an oblique direction. The convolution direction of the de-framing is the same as that of the pressing frame, and the convolution order of the de-framing is opposite to that of the pressing frame. If the frame-pressing only undergoes convolution in the vertical direction, the de-framing also only performs convolution in the vertical direction. Likewise, if the frame-pressing only undergoes convolution in the horizontal direction or oblique direction, the de-framing also only performs convolution in the horizontal direction or oblique direction. If the compressed frame undergoes convolution in multiple directions, the deframed convolution is also performed in multiple directions, and the direction and order of the deframed convolution are the same as those of the compressed frame. The direction and order of convolution are reversed. That is, the convolution in the vertical direction is performed first and then the convolution in the horizontal direction is performed in the frame pressing, and the convolution in the horizontal direction is performed first in the de-framing process, and then the convolution in the vertical direction is performed.

Step S340 may also include:

S346: Calculate the difference between the frame being decoded (that is, the decoded frame _{P 2} ) and the decoded spectrum adjustment frame PC to obtain a third boundary.

For the convenience of description, we define the data in the third boundary as P _E3 . The third boundary P _E3 is the boundary of the de-framing, which includes boundary information of the initial frame P ₀ .

In some embodiments, step S340 may further include:

_S347 : Enhance the third boundary PE3 by the second coefficient b.

Wherein, the second coefficient b is any number greater than 1. _In some embodiments, the third boundary _PE3 may be data obtained by calculating the difference between the decoded frame P2 and the decoded spectrum adjustment frame PC _. In other embodiments, the third boundary P _E3 may be a boundary enhanced by the second coefficient b. At this time, the third boundary P _E3 can also be expressed as the following formula:

P _E3 =P ₂ -P _C =P ₂ -P ₂ *H ₂ (f) Formula (6)

P _E3 =b*(P ₂ -P _C )=b*(P ₂ -P ₂ *H ₂ (f)) Formula (7)

As mentioned above, the components in the intermediate frequency to high frequency region in the decoded spectrum adjustment frame are filtered, and the difference between the decoded frame and the decoded spectrum adjustment frame can be obtained to obtain the intermediate frequency to high frequency in the decoded frame. component of the frequency region, that is, the boundary of the deframe. The deframing boundary includes boundary information of the initial frame. As before, the data in the deframed boundary is defined as the third boundary P _E3 . Wherein, b is an enhancement coefficient, which represents the enhancement degree of the boundary information, and the larger b is, the stronger the enhancement degree of the boundary information is. The adjustment coefficient b can be valued according to an empirical value, or can be obtained through machine learning training.

Step S340 may also include:

S348: Using the second gamma algorithm, weaken the boundary whose boundary value is within the second preset range in the third boundary P _E3 , so as to perform noise reduction on the third boundary P _E3 to obtain noise reduction boundary.

For the convenience of description, we define the data in the noise reduction boundary as P _E0 . The boundary correction includes attenuating the boundary whose boundary value is within the second preset range in the de-framed boundary through the second gamma algorithm, so as to perform noise reduction. In general, the image noise generated when encoding and decoding a video or image is usually within a small range. Therefore, when the data is decompressed, noise reduction processing can be performed on the image noise in this small range generated by encoding and decoding. When performing noise reduction processing, a gamma value γ greater than 1 may be used to weaken the boundary of the boundary value within the second preset range [-R4, R4], so as to achieve the effect of noise reduction. The second preset range [-R4, R4] may be a boundary value for which the boundary correction needs to be performed. The boundary value may be a value corresponding to each pixel in the third boundary _PE3 . R4 can be the boundary threshold. For example, R4 can be 30, 40, 50, etc. In some embodiments, R4 can be any number between 5-30.

FIG. 11B shows a flowchart of performing the boundary correction according to an embodiment of the present specification. As shown in FIG. 11B , step S348 may be: the data decompression device 300 adjusts the third boundary P _E3 through the third adjustment function H _L3 (f) to obtain a fourth boundary P _E4 ; The algorithm is to weaken the boundary in the fourth boundary P _E4 whose boundary value is within the second preset range [-R4, R4] to obtain the noise reduction boundary P _E0 .

In the third boundary _PE3 , the boundary that does not need to perform the boundary correction contains many components in the middle frequency to high frequency region. Therefore, in order to prevent the boundary correction from affecting other boundaries that do not need to be corrected, the data decompression device 300 may first filter the third boundary _PE3 to filter the components in the mid-frequency to high-frequency region. The third adjustment function H _L3 (f) may be a low-pass filter with a DC component DC equal to 1, so that the components in the low frequency region of the third boundary P _E3 in the frequency domain are preserved and the components in the intermediate frequency to high frequency region is filtered. The third adjustment function H _L3 (f) may be the same as the second adjustment function H _L2 (f).

The second gamma algorithm may be a gamma algorithm with a gamma value γ>1. In step S348, the data decompression apparatus 300 may perform boundary correction on the pixels in the fourth boundary _PE4 one by one. Specifically, the data decompression device 300 may compare the boundary value corresponding to each pixel in the fourth boundary PE4 with the boundary threshold R4; when the boundary value is within [ _-R4 , R4], the second boundary value is used. The gamma algorithm corrects the boundary value to make the absolute value of the boundary value smaller, so as to perform boundary noise reduction; when the boundary value is outside [-R4, R4], no correction is performed.

The data decompression device 300 may perform the boundary correction using a table look-up method. Table 5 may be stored in the data decompression device 300 . Table 5 stores the correspondence between the input boundary value and the output boundary value in the second gamma algorithm, that is, the correspondence between the boundary value before correction and the boundary value after correction. Taking R4=5 as an example, Table 5 can be expressed as:

FIG. 11C shows another flowchart of performing boundary correction according to an embodiment of the present specification. As shown in FIG. 11C , in step S348, the data decompression device 300 directly determines that the boundary value in the third boundary _PE3 is in the second preset range [−R5, The boundary within R5] is corrected to obtain the noise reduction boundary P _E0 . Wherein R5 can be different from R4 or can be the same as R4.

In some embodiments, the data decompression device 300 may further correct the boundary in the third boundary _PE3 within the second preset range [-R5, R5] by using the deviation value ΔE to obtain the obtained The noise reduction boundary P _E0 is described. Wherein, the noise reduction boundary _PE0 is the sum of the third boundary _PE3 and the deviation value ΔE. Table 6 may be stored in the data decompression device 300 . Table 6 stores the correspondence between the input boundary value of the third boundary _PE3 , the deviation value ΔE, and the input boundary value and the output boundary value. Taking R1=5 as an example, Table 6 can be expressed as:

The data decompression device 300 can directly obtain the noise reduction boundary 0 _E0 according to Table 6 .

S349: Superimpose the noise reduction boundary P _E0 and the decompressed frame (ie, the decoded frame P ₂ ) to obtain the decompressed frame P ₄ .

Since the boundary correction is only performed for the boundaries within a small range, the boundaries and non-boundary areas outside the second preset range are not corrected. Therefore, the decompressed frame P4 can be expressed as the following _formula :

P ₄ =P ₂ +P _E0

=P ₂ +P _E3 +ΔE

≈P′ ₀ ·H ₁ (f)·(1+b(1-H ₂ (f)))

≈P ₀ ·H ₁ (f)·(1+b(1-H ₂ (f))) Equation (8)

Taking video data as an example, since the human eye is sensitive to the information in the low-frequency to mid-frequency region, the design of H ₁ (f) only attenuates the amplitude of the low-frequency to medium-frequency region in the initial frame, and makes the coding spectrum adjust the frame. The frequency information of all frequencies from the low frequency to the intermediate frequency in the initial frame is retained; the data P ₂ in the decoded frame is basically the same as the data P ₁ in the encoded spectrum adjustment frame, so the low frequency to the intermediate frequency are also reserved in the decoded frame. the frequency information of the region; and the components from the intermediate frequency to the high frequency region in the decoded spectrum modulation frame are filtered, so the frequency information of the low frequency region is retained; therefore, the difference between the decoded frame and the decoded spectrum modulation frame is obtained in the decoded frame. The frequency information of the intermediate frequency region in the initial frame is retained in the boundary; and the noise reduction boundary only removes the noise in a small range in the boundary of the de-framed, and the boundary outside the second preset range No processing is performed; therefore, theoretically, without considering the deviation caused by other algorithms, the decompressed frame obtained by the superposition of the decoded frame and the noise reduction boundary can completely restore or basically restore all frequencies from the low frequency to the intermediate frequency in the initial frame. informational. That is to say, the data decompression can restore or even enhance the data compressed at any frequency from low frequency to intermediate frequency. Therefore, after data decompression, the amplitude of the decompressed frame at any frequency in the low frequency to medium frequency region should be approximately equal to or greater than that of the initial frame. The approximately equal means that the amplitude of the decompressed frame is equal to the amplitude of the initial frame, and fluctuates within a certain error range. Taking video data as an example, when the amplitude of the decompressed frame at any frequency in the low-frequency to mid-frequency region is restored to 85% or more of the initial frame, it is difficult for the human eye to perceive the difference between the decompressed frame and the original frame. the difference of the initial frame. Therefore, after data decompression, the amplitude of the decompressed frame at any frequency in the low-frequency to medium-frequency region should not be less than 85% of the initial frame. That is, the error range should not make the amplitude of the decompressed frame at any frequency in the low frequency to medium frequency region lower than 85% of the initial frame. The human eye is relatively insensitive to the information in the high frequency region, therefore, the information in the high frequency region in the decompressed frame can be retained to adapt to high-quality scenes, and can also be attenuated to suppress unnecessary high-frequency noise. The relationship between P ₀ and P ₄ can be expressed as the following formula:

或者

or

It should be noted that a certain range of errors can be allowed in the formula. For example, P ₄ ≥ P ₀ may be the basic value of P _4. In the case that P 4 is greater than or equal to P ₀ , P ₄ is allowed to fluctuate within a certain error range. That is, when P ₄ =P ₀ , P ₄ can allow P ₄ to be slightly smaller than P ₀ in the case of a negative error. The formula here only lists the basic relationship between P ₄ and P ₀ , and the error is not written into the formula. Those skilled in the art should understand that the fluctuation within the error range makes the amplitude of the decompressed frame in the low-frequency to medium-frequency region slightly smaller than The situation of the initial frame also falls within the protection scope of this specification. In the following formula, a certain range of error is also allowed. In the following, only a description is given of the basic relationship that the magnitude of P ₄ is greater than or equal to the initial frame P ₀ . The fluctuation within the error range can be deduced by those skilled in the art.

For the convenience of description, we define the overall spectrum adjustment function between P ₀ and P ₄ as H ₀ (f), then the relationship between P ₀ and P ₄ can be expressed as the following formula:

P ₄ =H ₀ (f)·P ₀ Formula (11)

Then, the overall spectrum adjustment function H ₀ (f) can be expressed as the following formula:

或者

or

Among them, f ₀ is the threshold value of the sensitive frequency of the human eye. For video data, f ₀ may be 0.33, or may be other values larger or smaller than 0.33. The value of _f0 is different for different types of data.

In H ₀ (f) in the above formulas (12) to (13), when H ₀ (f)≈1 in the selected frequency domain interval, the decompressed frame can be in the selected frequency domain interval The data of the decompressed frame is restored to the initial frame; when H ₀ (f)>1 in the selected frequency domain interval, the data of the decompressed frame in the selected frequency domain interval can be enhanced, that is, the decompression The magnitude of the frame in the selected area is higher than the initial frame. For example, if the initial frame is a frame in the video, as long as H ₀ (f) is greater than 1 in the selected frequency domain interval, sharpness enhancement can be achieved. For the convenience of description, we define H ₀ (f)≈1 as the normal mode and H ₀ (f)>1 as the enhanced mode. In the following, we will take video data as an example to describe the overall spectrum adjustment function H ₀ (f) in detail.

FIG. 12A shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification. FIG. 12B shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification. FIG. 12C shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification. FIG. 12D shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification. FIG. 12E shows a graph of an overall adjustment function H ₀ (f) provided according to an embodiment of the present specification. As shown in FIGS. 12A to 12E , the horizontal axis is the normalized frequency f, and the vertical axis is the amplitude adjustment gain H ₀ of the overall spectrum adjustment function H ₀ (f). The curves in Figures 12A to 12E represent different overall spectral adjustment functions H ₀ (f). The normalized frequency maximum value of the horizontal axis is 0.5. The normalized frequency f of the horizontal axis can be divided into a low frequency region, a middle and low frequency region, a middle frequency region, a middle and high frequency region and a high frequency region. The frequencies between (0, a] belong to low frequencies; the frequencies between (a, b] belong to the middle and low frequencies; the frequencies between (b, c] belong to the middle frequencies; the frequencies between (c, d) belong to the middle and high frequencies; The frequencies between (d, 0.5] belong to high frequencies. The values of a, b, c, d, and e are described with reference to FIG. 8A , which will not be repeated here.

Since the human eye is more sensitive to the low-frequency to medium-frequency data in the video data than to the high-frequency data, after the data is decompressed, try to keep the information in the low-frequency to medium-frequency region of the decompressed frame relative to the initial frame not lost. , that is to say, the overall spectrum adjustment function H ₀ (f) should make the amplitude of the decompressed frame in the low-frequency to mid-frequency region not less than 85% of the initial frame, and may even be larger than the initial frame. Since the human eye is not sensitive to the information in the high-frequency region, the amplitude of the decompressed frame in the high-frequency region can be selected according to different application scenarios. The amplitude of the frequency region may be smaller than the initial frame. In a reconnaissance scenario that requires high definition, the amplitude of the decompressed frame in the high frequency region may be approximately equal to or greater than the initial frame. As shown in FIGS. 12A to 12E , the amplitude adjustment gain H ₀ of the overall adjustment function H ₀ (f) at any frequency f in the low-frequency to intermediate-frequency region (including the low-frequency and intermediate-frequency regions) is greater than 1 or approximately equal to 1, so that after decompression The amplitude of the decompressed frame is not less than 85% of the initial frame, so that the definition can be restored or enhanced, and the visual observation effect can be improved. Said approximately equal to 1 may be fluctuated within a certain error range of equal to 1 here. The error range can be 0, ±1%, ±2%, ±3%, ±4%, ±5%, ±6%, ±7%, ±8%, ±9%, ±10%, ±11 %, ±12%, ±13%, ±14%, ±15%, etc., within the interval specified by any two values. For the convenience of description, we define the amplitude adjustment gain of the overall adjustment function H ₀ (f) in the high frequency region as the first amplitude adjustment gain, and the amplitude adjustment gain in the intermediate frequency region as the second amplitude adjustment gain. The amplitude adjustment gain in the low frequency region is defined as the third amplitude adjustment gain. The third amplitude adjustment gain value, the second amplitude adjustment gain value and the first amplitude adjustment gain value may fluctuate within the error range.

As shown in FIG. 12A , the third amplitude adjustment gain value, the second amplitude adjustment gain value and the first amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the low frequency to high frequency region are all approximately equal to 1, so that The amplitude of the decompressed frame in the low frequency to high frequency region is not less than 85% of the initial frame, so that the data of the decompressed frame in the low frequency to high frequency region can be restored smoothly or basically to the state of the initial frame.

As shown in FIG. 12B , the third amplitude adjustment gain value and the second amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the low frequency to medium frequency region are approximately equal to 1, so that the decompression frame is in the low frequency to medium frequency region. The data can be restored smoothly or basically to the state of the original frame. The first amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the high-frequency region is less than 1, so that the amplitude of the decompressed frame in the high-frequency region is smoothly reduced relative to the initial frame, so as to suppress high-frequency noise . The steady decrease of the amplitude value may be that the amplitude value is attenuated at the first amplitude value adjustment gain value, or the amplitude value may be attenuated within a certain error range near the first amplitude value adjustment gain value. For example, the first amplitude adjustment gain may be any value between 0 and 1. For example, the first amplitude adjustment gain value may be 0, 0.04, 0.08, 0.12, 0.16, 0.20, 0.24, 0.28, 0.32, 0.36, 0.40, 0.44, 0.48, 0.52, 0.56, 0.60, 0.64, 0.68, 0.72 , 0.76, 0.80, 0.84, 0.88, 0.92, 0.96 and 1 within the interval specified by any two values. As shown in FIG. 12B , the first amplitude adjustment gain of the overall adjustment function H ₀ (f) in the high frequency region (approximately the interval of 0.4 to 0.5) is about 0.6. Both the second and third amplitude adjustment gain values are around 1. The second and third amplitude adjustment gain values may fluctuate within a certain error range, for example, the second and third amplitude adjustment gain values may be in values such as 0.85, 0.90, 0.95, 1, 1.05, 1.10, and 1.15 within any two specified intervals.

As shown in FIG. 12C , the third amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the low frequency region is approximately equal to 1, so that the data of the decompressed frame in the low frequency region can be restored smoothly or basically restored to the original frame. state. The second amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the intermediate frequency region and the first amplitude adjustment gain value in the high frequency region are both greater than 1, so that the amplitude of the decompressed frame in the intermediate frequency to high frequency region is Increases smoothly relative to the initial frame, resulting in enhanced data clarity in the mid to high frequency region. The steady increase of the amplitude value may be that the amplitude value is enhanced by the second amplitude value adjustment gain value and the first amplitude value adjustment gain value, or the amplitude value may be increased between the second amplitude value adjustment gain value and the other amplitude value adjustment gain value. The enhancement is performed within a certain error range near the first amplitude adjustment gain value. The second amplitude adjustment gain value and the first amplitude adjustment gain value may be substantially the same, or the second amplitude adjustment gain value may be greater than the first amplitude adjustment gain value, or the The second amplitude adjustment gain value is smaller than the first amplitude adjustment gain value. In the curve shown in FIG. 12C , the second amplitude adjustment gain value and the first amplitude adjustment gain value are substantially the same. The second amplitude adjustment gain value and the first amplitude adjustment gain value may be any value greater than 1. For example, the second amplitude adjustment gain value and the first amplitude adjustment gain value may be at 1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, 2.1, 2.2 and 2.4 Within the interval specified by any two of the equivalent values. As shown in FIG. 12C , the second amplitude adjustment gain and the first amplitude adjustment gain of the overall adjustment function H ₀ (f) in the mid-frequency to high-frequency region are around 1.2.

As shown in FIG. 12D , the third amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the low frequency region is approximately equal to 1, so that the data of the decompressed frame in the low frequency region can be restored smoothly or basically to the original frame. state. The second amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the intermediate frequency region is greater than 1, so that the amplitude of the decompressed frame in the intermediate frequency increases smoothly relative to the initial frame, so that the data clarity in the intermediate frequency region is improved. enhanced. The first amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the high frequency region is less than 1, so that the amplitude of the decompressed frame in the high frequency region is smoothly reduced relative to the initial frame, so that the insensitive The amount of data in the high frequency region is reduced to suppress high frequency noise. The curve shown in Figure 12D enhances clarity while reducing the amount of data. The second amplitude adjustment gain value may be any value greater than 1. The first amplitude adjustment gain may be any value between 0 and 1. As shown in FIG. 12D , the second amplitude adjustment gain of the overall adjustment function H ₀ (f) in the intermediate frequency region is about 1.2, and the first amplitude adjustment gain in the high frequency region is about 0.6.

As shown in FIG. 12E , the third amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the low frequency region is greater than 1, so that the amplitude of the decompressed frame in the low frequency region increases smoothly relative to the initial frame. The second amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the intermediate frequency region is greater than 1, so that the amplitude of the decompressed frame in the intermediate frequency region is steadily increased relative to the initial frame, so as to make the low frequency to the intermediate frequency region. Data clarity is enhanced. Wherein, the second amplitude adjustment gain value may be equal to the third amplitude adjustment gain value, or may be greater than the third amplitude adjustment gain value. In the curve shown in FIG. 12E , the second amplitude adjustment gain value is greater than the third amplitude adjustment gain value, so that the amplitude of the decompressed frame in the intermediate frequency region is increased by a larger amplitude than that in the low frequency region. , so as to enhance the clarity of the intermediate frequency region where the human eye is most sensitive and improve the visual observation effect. The first amplitude adjustment gain value of the overall adjustment function H ₀ (f) in the high frequency region is less than 1, so that the amplitude of the decompressed frame in the high frequency region is smoothly reduced relative to the initial frame, so that the insensitive The amount of data in the high frequency region is reduced to suppress high frequency noise. The curve shown in Figure 12E enhances clarity while reducing the amount of data. The third amplitude adjustment gain value may be a value slightly greater than 1. For example, the third amplitude adjustment gain value may be within an interval specified by any two values of 1, 1.04, 1.08, 1.12, 1.16, and 1.2. The second amplitude adjustment gain value may be any value greater than the third amplitude adjustment gain. For example, the second amplitude adjustment gain value and the first amplitude adjustment gain value may be at 1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, 2.1, 2.2 and 2.4 Within the interval specified by any two of the equivalent values. The first amplitude adjustment gain may be any value between 0 and 1. For example, the first amplitude adjustment gain value may be 0, 0.04, 0.08, 0.12, 0.16, 0.20, 0.24, 0.28, 0.32, 0.36, 0.40, 0.44, 0.48, 0.52, 0.56, 0.60, 0.64, 0.68, 0.72 , 0.76, 0.80, 0.84, 0.88, 0.92, 0.96 and 1 within the interval specified by any two of the values. As shown in FIG. 12E , the third amplitude adjustment gain of the overall adjustment function H ₀ (f) in the low frequency region is about 1.1, the second amplitude adjustment gain in the intermediate frequency region is about 1.2, and the third amplitude adjustment gain in the high frequency region is about 1.2. A value adjustment gain is around 0.6.

Further, when the high frequency region is not connected with the intermediate frequency region, the overall spectrum adjustment function H ₀ (f) can also be adjusted in the amplitude of the high frequency region, so that the amplitude adjustment gain Changes in the mid-high frequency region are smooth and continuous.

Further, when the intermediate frequency region is not connected with the low frequency region, the overall spectrum adjustment function H ₀ (f) can also be adjusted in the amplitude of the intermediate and low frequency region, so that the amplitude adjustment gain is at Changes in the mid-low frequency region are continuous.

The curve of the overall adjustment function H ₀ (f) is a smooth transition curve. In terms of engineering implementation, on the basis of realizing that the amplitude of the decompressed frame in the low frequency to medium frequency region is approximately equal to or greater than that of the initial frame, a small range of fluctuations in the curve of the overall adjustment function H ₀ (f) can be allowed, so The above fluctuation does not affect the effect of decompression. For data in other forms than video data, the parameters of the overall adjustment function H ₀ (f) can be set according to the sensitivity of the receiver to the data. Different forms of data have different degrees of frequency sensitivity of receivers.

For the convenience of description, we will take the case shown in formula (13) as an example for description. Combining formula (8) and formula (13), the decompressed frame P ₄ can be expressed as the following formula:

At this time, the relationship between the encoding spectrum adjustment function H ₁ (f) corresponding to the encoding convolution kernel and the decoding spectrum adjustment function H ₂ (f) corresponding to the decoding convolution kernel can be expressed as the following formula:

Therefore, the relationship between H ₁ (f) and H ₂ (f) can be expressed as the following formula:

Among them, in the decoding spectrum adjustment function H ₂ (f), except that the amplitude adjustment gain of the part whose frequency is 0 is 1, the amplitude adjustment gain of other frequencies is less than 1, so 1/(1+b(1 -H ₂ (f))) is less than 1 at other frequencies except frequency 0, so formula (16) can guarantee the amplitude of the part of frequency 0 in the encoding spectrum adjustment function H ₁ (f) The adjustment gain is 1, while the amplitude adjustment gain corresponding to other frequencies is less than 1.

As mentioned above, if the initial frame undergoes convolution in multiple directions, the decoded frame also undergoes convolution in multiple directions, and the decoded frame is convolved in direction and order with the initial frame The direction and order are reversed during convolution. That is, the initial frame first performs convolution in the vertical direction and then performs convolution in the horizontal direction, and the decoded frame performs convolution in the horizontal direction first and then performs convolution in the vertical direction. It should be noted that the decoded frame needs to perform horizontal convolution to obtain the compensation information in the horizontal direction, and after superimposing the compensation information in the horizontal direction of the decoded frame with the decoded frame, then perform convolution in the vertical direction to obtain the compensation information. compensation information in the vertical direction, and superimpose the compensation information in the vertical direction of the decoded frame with the decoded frame.

或者编码频谱调节函数乘积组合

或者线性组合和乘积组合的组合都属于本说明书保护的范围。其中，i≥1，

代表n个函数的线性组合，H_2i(f)代表第i个函数，k_i代表第i个函数对应的权重。j≥1，

代表n个函数的乘积组合，k_j代表第j个函数对应的权重，H_2j(f)可以是任意函数。13A shows a graph of an overall adjustment function H ₀ (f), an encoding spectrum adjustment function H ₁ (f), and a decoded spectrum adjustment function H ₂ (f) in the normal mode provided according to an embodiment of the present specification. FIG. 13B shows graphs of the overall adjustment function H ₀ (f), the encoding spectrum adjustment function H ₁ (f) and the decoded spectrum adjustment function H ₂ (f) of an enhancement mode provided according to an embodiment of the present specification. The encoding convolution kernel and the decoding convolution kernel used in FIG. 13A and FIG. 13B are the same, and the adjustment coefficient b is the same. In FIGS. 13A and 13B, b=1.5 is taken as an example for description. As shown in FIGS. 13A and 13B , the horizontal axis is the normalized frequency f, and the vertical axis is the amplitude adjustment gain H. As shown in FIG. 13A , the overall spectrum adjustment function H ₀ (f)≈1 in any frequency region, the overall spectrum adjustment function H ₀ (f) performs the normal mode spectrum adjustment on the superimposed decompression frame, that is, the overall spectrum adjustment function The information for all frequencies in H ₀ (f) is completely preserved, and the data in the decompressed frame can be basically restored to the data in the initial frame. As shown in FIG. 13B , the overall spectral adjustment function H ₀ (f)≈1 in the low frequency region, and the overall spectral adjustment function H ₀ (f)>1 in the middle frequency to high frequency region. The overall spectrum adjustment function H ₀ (f) performs spectrum adjustment in the enhancement mode on the intermediate frequency to high frequency region of the decompressed frame, that is, the information in the intermediate frequency to high frequency region in the overall spectrum adjustment function H ₀ (f) is enhanced, so The data in the intermediate frequency to high frequency region in the decompressed frame is enhanced compared to the data in the intermediate frequency to high frequency region in the initial frame. It should be noted that the curves shown in FIG. 13A and FIG. 13B are only illustrative, and those skilled in the art should understand that the curves of H ₀ (f), H ₁ (f) and H ₂ (f) are not limited to In the forms shown in FIGS. 13A and 13B , all H ₀ (f), H ₁ (f), and H ₂ (f) curves conforming to formula (15) fall within the protection scope of this specification. It should be pointed out that all the decoding spectrum adjustment functions conforming to formula (15) are linearly combined

Or code spectrum adjustment function product combination

Or the combination of linear combination and product combination all belong to the protection scope of this specification. Among them, i≥1,

Represents the linear combination of n functions, H _2i (f) represents the _ith function, and ki represents the weight corresponding to the ith function. j≥1,

Represents the product combination of n functions, k _j represents the weight corresponding to the jth function, and H _2j (f) can be any function.

Table 7 shows a parameter table of a decoding convolution kernel provided according to an embodiment of the present specification. Table 7 exemplarily lists the parameters of a decoding convolution kernel. The parameters of the decoding convolution kernel are all non-negative numbers, so that the data convolved by the decoding convolution kernel avoids the ringing effect. Table 7 is only an exemplary illustration, and those skilled in the art should know that the decoding convolution kernel is not limited to the parameters shown in Table 7, and all decoding convolution kernels that meet the foregoing requirements belong to the protection scope of this specification.

Table 8 shows a parameter table of a normal mode encoding convolution kernel provided according to an embodiment of the present specification. Table 8 exemplarily lists the parameters of a normal mode encoding convolution kernel. The encoding convolution kernel of the normal mode is based on the overall spectrum adjustment function H ₀ (f) of the normal mode, and the decoding spectrum adjustment function H ₂ (f) corresponding to the parameter table of the decoding convolution kernel shown in Table 7. ), the obtained coded spectrum adjustment function H ₁ (f) is obtained by Fourier transform. That is, the encoding spectrum adjustment function H ₁ (f) is obtained corresponding to H ₀ (f)=1. The data compression apparatus 200 and the data decompression apparatus 300 can use the normal mode encoding convolution kernel shown in Table 8 and the decoding convolution kernel shown in Table 7 to make the data of the superimposed frame and the initial The data of the frame is basically the same. Table 8 is only an exemplary illustration. Those skilled in the art should know that the encoding convolution kernel of the normal mode is not limited to the parameters shown in Table 8, and all encoding convolution kernels that meet the aforementioned requirements belong to the protection of this specification. scope.

Table 9 shows a parameter table of a coding convolution kernel of an enhancement mode provided according to an embodiment of the present specification. The encoding convolution kernel of the enhancement mode is based on the overall spectrum adjustment function H ₀ (f) of the enhancement mode, and the decoding spectrum adjustment function H ₂ (f) corresponding to the parameter table of the decoding convolution kernel shown in Table 7. ), the obtained coded spectrum adjustment function H ₁ (f) is obtained by Fourier transform. That is, the encoding spectrum adjustment function H ₁ (f) is obtained corresponding to H ₀ (f)>1. The data compression apparatus 200 can enhance the data of the superimposed frame using the encoding convolution kernel of the enhancement mode shown in Table 9 and the decoding convolution kernel shown in Table 7. Table 9 is only an exemplary illustration. Those skilled in the art should know that the encoding convolution kernel of the enhanced mode is not limited to the parameters shown in Table 9, and all encoding convolution kernels that meet the aforementioned requirements belong to the protection of this specification. scope.

It should be noted that, after the convolution operation, a normalization process needs to be performed, so that the grayscale value of the image after the convolution operation is between 0 and 255.

It should be noted that, when the code rate is high, the noise in the decompressed frame is very small, and the above boundary correction may not be performed. In the case of a low bit rate, or in the enhancement mode, that is, the mode in which H ₀ (f)>1, excessive enhancement may cause noise in the decompressed frame, affecting the effect of visual observation. We can perform the boundary correction on the boundary of the decompressed frame to obtain the decompressed frame, so as to effectively eliminate noise.

Fig. 14A shows an example diagram of a decompressed image without boundary correction provided according to an embodiment of this specification; Fig. 14B shows a decompressed image with boundary correction provided according to an embodiment of this specification sample graph. Comparing FIG. 14A and FIG. 14B , it is found that the boundary correction method described in this specification can effectively eliminate noise.

To sum up, the data processing system 100 provided in this specification, when compressing the initial data, executes the method P200 by the data compression device 200 to compress the initial data in the small range of the initial frame The boundary is adjusted by the first gamma algorithm to avoid the loss of boundary information with a small difference between adjacent pixel values during the data compression process and the loss of details, while the method P200 uses coding convolution for the initial frame. The kernel performs encoding spectrum adjustment, so that the amplitude of the initial frame in the low-frequency to high-frequency region in the frequency domain is steadily reduced, thereby reducing the data information in the initial frame, improving the encoding efficiency, and reducing the compressed data capacity. , improve data compression efficiency and data transmission efficiency. The data processing system 100 provided in this specification, when decompressing the compressed frame, executes the method P300 through the data decompression device 300, uses a decoding convolution kernel to perform decoding spectrum adjustment on the compressed frame, and uses a smooth transition decoding spectrum The adjustment function H ₂ (f) performs spectrum adjustment on the compressed frame, filters the components in the intermediate frequency to high frequency region in the compressed frame, and then calculates the difference between the compressed frame and the compressed frame adjusted by the decoded spectrum, The boundary information of the initial frame is obtained, and the boundary within a small range in the boundary information of the initial frame is weakened by the second gamma algorithm to eliminate the noise in the boundary; The boundary information is superimposed to obtain the decompressed frame. Wherein, the decoding convolution kernel corresponding to the decoding spectrum adjustment function H ₂ (f) corresponds to the encoding convolution kernel, and all coefficients are non-negative numbers, or the ratio of the absolute value of the sum of the negative coefficients to the sum of the non-negative coefficients is less than 0.1, thereby effectively avoiding the appearance of noise and making the decompressed frame clearer. The method and system can improve data compression efficiency, improve transmission efficiency, avoid loss of details, and at the same time can improve the clarity of decompressed data and effectively eliminate noise.

This specification further provides a non-transitory storage medium storing at least one set of executable instructions for data processing, and when the executable instructions are executed by a processor, the executable instructions instruct the processor to implement The steps of data processing method P200. In some possible implementations, various aspects of this specification may also be implemented in the form of a program product, which includes program code. When the program product is run on the data compression device 200, the program code is used to cause the data compression device 200 to perform the data processing steps described in this specification. A program product for implementing the above-described method may employ a portable compact disc read only memory (CD-ROM) and include program code, and may run on a data compression device 200, such as a personal computer. However, the program product of this specification is not limited thereto, and in this specification, a readable storage medium may be any tangible medium that contains or stores a program that can be used by or combined with an instruction execution system (eg, the compressor side processor 220 ). use. The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of readable storage media include: electrical connections with one or more wires, portable disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory ( EPROM or flash memory), optical fiber, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. The computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code therein. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A readable storage medium can also be any readable medium other than a readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing. Program code for carrying out the operations of this specification may be written in any combination of one or more programming languages, including object-oriented programming languages—such as Java, C++, etc., as well as conventional procedural Programming Language - such as the "C" language or similar programming language. The program code may execute entirely on data compression device 200, partly on data compression device 200, as a stand-alone software package, partly on data compression device 200 and partly on a remote computing device, or entirely remotely Execute on the computing device. In cases involving a remote computing device, the remote computing device may be connected to data compression device 200 through transmission medium 120, or, alternatively, may be connected to an external computing device.

The foregoing describes specific embodiments of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that in the embodiments and still achieve desirable results. Additionally, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

In conclusion, after reading this detailed disclosure, those skilled in the art will appreciate that the foregoing detailed disclosure may be presented by way of example only, and may not be limiting. Although not explicitly described herein, it will be understood by those skilled in the art that this description needs to encompass various reasonable changes, improvements and modifications to the embodiments. Such changes, improvements and modifications are intended to be suggested by this specification and are within the spirit and scope of the exemplary embodiments of this specification.

Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, "one embodiment," "an embodiment," and/or "some embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of this specification. Thus, it is emphasized and should be understood that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various parts of this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures or characteristics may be combined as appropriate in one or more embodiments of this specification.

It will be appreciated that, in the foregoing description of embodiments of this specification, in order to aid in the understanding of one feature, the specification, for the purpose of simplifying the specification, groups various features in a single embodiment, drawings, or description thereof. However, this does not mean that the combination of these features is necessary, and it is entirely possible for those skilled in the art to extract some of the features as a separate embodiment to understand when reading this specification. That is to say, the embodiments in this specification can also be understood as the integration of multiple sub-embodiments. It is also true that each sub-embodiment contains less than all features of a single foregoing disclosed embodiment.

Each patent, patent application, publication of a patent application, and other materials, such as articles, books, specifications, publications, documents, articles, etc., cited herein, may be incorporated herein by reference. For all purposes in its entirety, except any filing history with which it relates, any identical filing that may be inconsistent or conflicting with this document, or any identical filing that may have a restrictive effect on the broadest scope of the claims history. associated with this document now or in the future. For example, in the event of any inconsistency or conflict between the descriptions, definitions, and/or use of terms, descriptions, definitions, and/or terms associated with any of the included materials, use The term shall prevail.

Finally, it should be understood that the embodiments of the application disclosed herein are illustrative of the principles of the embodiments of the present specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are merely illustrative and not limiting. Those skilled in the art may adopt alternative configurations according to the embodiments in this specification to implement the applications in this specification. Accordingly, the embodiments of this specification are not limited to those precisely described in the application.

Claims (25)

1. A method of data processing, comprising:

selecting an initial frame in the initial data, wherein the initial frame comprises initial data with preset byte number; and

performing data compression on the initial frame to obtain a compressed frame, wherein the data compression includes performing boundary adjustment on the initial frame and performing coding spectrum adjustment on a compressed frame, the compressed frame includes the initial frame and any data state before the initial frame becomes the compressed frame in the data compression process,

wherein the encoded spectral conditioning includes convolving the intra-compressed frame with an encoded convolution kernel to smoothly reduce the amplitude of the intermediate frequency region of the intra-compressed frame in the frequency domain.

2. The method of data processing as claimed in claim 1, wherein the boundary adjustment includes adjusting a boundary having a boundary value within a first preset range among the boundaries of the initial frame by a first gamma algorithm having a gamma value less than 1.

3. The method of data processing according to claim 2, wherein said data compressing said initial frame comprises:

firstly, the boundary adjustment is carried out on the initial frame, and then the coding spectrum adjustment is carried out; or alternatively

And firstly, carrying out the coding spectrum adjustment on the initial frame, and then carrying out the boundary adjustment.

4. The method of data processing according to claim 3, wherein said boundary adjustment prior to said code spectrum adjustment for said initial frame comprises:

performing the boundary adjustment on the initial frame to obtain a first enhanced frame; and

performing the coded spectral modification and coding on the first enhancement frame, including one of:

performing the coding spectrum adjustment on the first enhancement frame, and then performing prediction and residual calculation on the first enhancement frame after the coding spectrum adjustment, wherein the compressed frame comprises the first enhancement frame;

predicting the first enhancement frame to obtain a prediction frame, and then performing the coding spectrum adjustment and residual calculation on the first enhancement frame and the prediction frame, wherein the on-press frame comprises the first enhancement frame and the prediction frame; and

and predicting and solving a residual error of the first enhanced initial frame, and then performing the coded spectrum regulation on the residual error, wherein the compressed frame comprises the residual error.

5. The method of data processing according to claim 4, wherein said performing said boundary adjustment on said initial frame to obtain a first enhanced frame comprises:

adjusting the initial frame through a first adjusting function to obtain a first frame, so that the component of the initial frame in a low-frequency region in a frequency domain is reserved and the component of a middle-frequency region to a high-frequency region is attenuated;

obtaining a first boundary by subtracting the initial frame and the first frame, wherein the first boundary comprises boundary information of the initial frame;

adjusting the boundary of which the boundary value is within the first preset range in the first boundary through the first gamma algorithm to obtain an enhanced boundary; and

and superposing the first frame and the enhanced boundary to obtain the first enhanced frame.

6. The method of data processing according to claim 5, wherein said performing said boundary adjustment on said initial frame to obtain a first enhanced frame prior to said obtaining an enhanced boundary, further comprises:

and enhancing the first boundary by a first coefficient, wherein the first coefficient is an arbitrary number greater than 1, and the first boundary comprises the boundary enhanced by the first coefficient.

7. The data processing method of claim 5, wherein the adjusting the boundary of the first boundary whose boundary value is within the first preset range to obtain an enhanced boundary comprises:

adjusting the first boundary through a second adjusting function to obtain a second boundary, so that the component of the first boundary in a low-frequency region in a frequency domain is reserved and the component of a medium-frequency region to a high-frequency region is attenuated; and

and adjusting the boundary of which the boundary value is within the first preset range in the second boundary through the first gamma algorithm to obtain the enhanced boundary.

8. The method of data processing according to claim 3, wherein said performing said code spectrum adjustment prior to said boundary adjustment for said initial frame comprises:

performing the coding spectrum adjustment on the initial frame to obtain a coding spectrum adjustment frame;

performing the boundary adjustment on the coding frequency spectrum adjustment frame to obtain a second enhancement frame; and

and predicting and residual solving are carried out on the second enhanced frame.

9. The data processing method of claim 8, wherein said performing said boundary adjustment on said encoded spectrally adjusted frame to obtain a second enhancement frame comprises:

obtaining a first boundary by subtracting the initial frame and the coding spectrum adjustment frame, wherein the first boundary comprises boundary information of the initial frame;

adjusting the boundary of which the boundary value is within the first preset range in the first boundary through the first gamma algorithm to obtain an enhanced boundary;

obtaining a first boundary of the first image, and obtaining an enhanced boundary; and

and superposing the coding frequency spectrum adjusting frame and the adjusting value to obtain the second enhancement frame.

10. The data processing method of claim 9, wherein said boundary adjusting said coded spectral adjustment frame prior to said deriving an enhancement boundary, resulting in a second enhancement frame, further comprising:

and enhancing the first boundary by a first coefficient, wherein the first coefficient is an arbitrary number greater than 1, and the first boundary comprises the boundary enhanced by the first coefficient.

11. The data processing method of claim 9, wherein the adjusting the boundary of the first boundary whose boundary value is within the first predetermined range to obtain an enhanced boundary comprises:

adjusting the first boundary through a second adjusting function to obtain a second boundary, so that the component of the first boundary in a low-frequency area in a frequency domain is reserved, and the component of a medium-frequency area to a high-frequency area is attenuated; and

and adjusting the boundary of which the boundary value is within the first preset range in the second boundary through the first gamma algorithm to obtain the enhanced boundary.

12. The method of data processing according to claim 1, wherein said code spectrum adjustment of said on-press frame comprises:

determining a frame type of the initial frame, the frame type comprising at least one of an intra-predicted frame, a forward-predicted frame, and a bi-directionally predicted frame; and

and selecting one convolution kernel from the coding convolution kernel group as the coding convolution kernel to convolute the compressed frame based on the frame type of the initial frame.

13. The method of data processing according to claim 12, wherein said convolving the compressed frame comprises:

and performing convolution on the compressed frame in at least one direction of a vertical direction, a horizontal direction and an oblique direction.

14. The method of data processing according to claim 1, wherein said encoded spectral modification is such that the amplitude of said at-compressed-frame high-frequency region is reduced smoothly in the frequency domain.

15. The method of data processing according to claim 1, wherein said encoding spectral adjustment is such that the amplitude of the low frequency region of the compressed frame is reduced smoothly in the frequency domain, and

the encoded spectral modification reduces the amplitude of the low frequency region of the compressed frame by a lower amplitude than the amplitude of the mid frequency region.

16. The method of data processing according to claim 1, wherein said encoded spectral modification provides a gain greater than zero for amplitude modification of said compressed frame at any frequency in the frequency domain.

17. A system for data processing, comprising:

at least one storage medium storing at least one set of instructions for data processing; and

at least one processor communicatively coupled to the at least one storage medium,

wherein when the system is running, the at least one processor reads the at least one instruction set and performs the method of data processing of any of claims 1-16 in accordance with the instructions of the at least one instruction set.

18. A method of data processing, comprising:

obtaining compressed data, wherein the compressed data comprises a compressed frame obtained by performing data compression on an initial frame; and

decompressing the compressed frame to obtain a decompressed frame, the decompressing includes performing decoding spectral adjustment and boundary correction on a decoded frame, the decoded frame includes the compressed frame and the compressed frame becomes any data state before the decompressed frame in the data decompressing process, wherein,

said decoding spectral modification includes convolving said decoded spectral modification with a decoding convolution kernel so that said decoded spectral modification is reduced smoothly in frequency domain in order to filter components from the intermediate frequency to the high frequency region, said encoded spectral modification having a predetermined relationship with said decoded spectral modification,

the boundary correction includes correcting, by a second gamma algorithm, a boundary of which a boundary value is within a second preset range among the boundaries of the decoded frame to perform noise reduction.

19. The data processing method of claim 18, wherein the second gamma algorithm includes a gamma algorithm having a gamma value greater than 1 to attenuate a boundary of the boundary value within the second preset range among the boundaries of the solution frame.

20. The method of data processing according to claim 18, wherein said data decompressing the compressed frame comprises:

performing the decoding frequency spectrum adjustment on the decoding frame to obtain a decoding frequency spectrum adjustment frame;

obtaining a third boundary by subtracting the decoding spectrum adjustment frame from the decoding frame, wherein the third boundary is the boundary of the decoding frame and comprises the boundary information of the initial frame;

weakening the boundary of which the boundary value is within the second preset range in the third boundary through the second gamma algorithm so as to reduce noise of the third boundary to obtain a noise reduction boundary; and

and superposing the denoising boundary and the decoding frame to obtain the decompression frame.

21. The method of data processing according to claim 20, wherein said decompressing said compressed frame prior to said deriving said de-noised boundary comprises:

and enhancing the third boundary by a second coefficient, wherein the second coefficient is an arbitrary number greater than 1, and the third boundary comprises a boundary enhanced by the second coefficient.

22. The data processing method of claim 20, wherein the attenuating, by the second gamma algorithm, the boundary of the third boundary whose boundary value is within the second preset range to denoise the third boundary to obtain a denoised boundary comprises:

adjusting the third boundary through a third adjusting function to obtain a fourth boundary, so that the component of the third boundary in a low-frequency region in a frequency domain is reserved, and the component from a medium-frequency region to a high-frequency region is filtered; and

and weakening the boundary of which the boundary value is within the second preset range in the fourth boundary through the second gamma algorithm.

23. The method of data processing according to claim 20, wherein said decompressing said compressed frame prior to said performing said decoded spectral conditioning on said frame under decoding to obtain a decoded spectral conditioned frame, further comprises:

and decoding the compressed frame to obtain a decoded frame, wherein the decoded frame comprises the decoded frame.

24. The method of data processing according to claim 18, wherein said data compression includes code spectrum adjustment, including convolving said compressed frame with a code convolution kernel to smoothly reduce in a frequency domain an amplitude in an intermediate frequency region of the compressed frame, said compressed frame including said initial frame and any data state of said initial frame prior to said compressed frame being said compressed frame in said data compression process,

wherein the decoding convolution kernel corresponds to the encoding convolution kernel.

25. A system for data processing, comprising:

at least one storage medium storing at least one set of instructions for data processing; and

at least one processor communicatively coupled to the at least one storage medium,

wherein when the system is running, the at least one processor reads the at least one instruction set and performs the method of data processing according to the instructions of the at least one instruction set.

Images