CN118714329A - Image compression method and device, and readable storage medium - Google Patents

Abstract

The present invention provides an image compression method and device, and a readable storage medium, comprising: performing K-level wavelet transform on an input image to obtain a first wavelet coefficient distribution; determining a common coefficient N of a low-frequency region of the first wavelet coefficient distribution, the common coefficient N not exceeding the difference obtained by subtracting the minimum value of the low-frequency region from the maximum absolute value of other regions, and the other regions are all non-low-frequency regions in the first wavelet coefficient distribution; subtracting the common coefficient N from each value of the low-frequency region to obtain an updated low-frequency region; the updated low-frequency region and other regions constitute a second wavelet coefficient distribution; transmitting the common coefficient; and scanning and outputting the second wavelet coefficient distribution using an embedded zerotree wavelet coding algorithm. The present invention can improve image compression speed and image compression effect.

Description

Image compression method and device, and readable storage medium

Technical Field

The present invention relates to the field of image compression, and in particular to an image compression method and device, and a readable storage medium.

Background Art

Image compression facilitates efficient storage, management and transmission of information. Currently, discrete wavelet transform (DWT) and embedded zerotree wavelet coding (Embedded Zerotree Wavelet Coding, EZW) have been widely used in image compression processing.

DWT transfers image information from the time domain to the frequency domain for analysis, and obtains information of different spaces and frequencies by performing multi-resolution decomposition of the image. Since the coefficients after wavelet transform have certain characteristics, the coefficients of wavelet transform are divided into parent-child nodes, and then the EZW algorithm is used to continuously scan the image after wavelet transform to generate more zero trees to encode the image. The more zero tree roots are generated, the less data will be used to represent the image.

The combination of DWT and EZW can effectively improve the compression effect. However, in order to meet the growing requirements of mobile terminals with limited resources for image processing, the image compression effect needs to be further improved.

Summary of the invention

One of the purposes of the present invention is to overcome at least some of the deficiencies in the prior art and to provide an image compression method and device, and a readable storage medium.

The technical solution provided by the present invention is as follows:

An image compression method comprises: performing a K-level wavelet transform on an input image to obtain a first wavelet coefficient distribution;

Determine a common coefficient N of a low-frequency region of the first wavelet coefficient distribution, wherein the common coefficient N does not exceed a difference obtained by subtracting a minimum value of the low-frequency region from a maximum absolute value of other regions, wherein the other regions are all non-low-frequency regions in the first wavelet coefficient distribution;

Subtracting the common coefficient N from each value of the low-frequency region to obtain an updated low-frequency region; the updated low-frequency region and the other regions constitute a second wavelet coefficient distribution;

Transmission common coefficient N;

The second wavelet coefficient distribution is encoded using an embedded zerotree wavelet coding algorithm, and the encoding result is transmitted.

In some embodiments, the common coefficient N is equal to the difference between the minimum value in the low frequency region and the maximum absolute value in other regions.

In some embodiments, for an input image of 2 ^L ×2 ^M , M ≥ L ≥ 3, set K = L - 1.

In some embodiments, for an input image of 2 ^L ×2 ^M , M≥L≥3, set K=Lm, m>1.

In some embodiments, it includes: during decoding, after restoring the second wavelet coefficient distribution, corresponding compensation is performed on each value in the low-frequency region of the second wavelet coefficient distribution according to the received common coefficients.

The present invention also provides an image compression device, comprising:

A wavelet transform module, used for performing a K-level wavelet transform on the input image to obtain a first wavelet coefficient distribution;

A common coefficient determination module, used to determine a common coefficient N of a low-frequency region of the first wavelet coefficient distribution, wherein the common coefficient N does not exceed a difference obtained by subtracting a maximum absolute value of other regions from a minimum value of the low-frequency region, wherein the other regions are all non-low-frequency regions in the first wavelet coefficient distribution;

A low frequency updating module, configured to subtract the common coefficient N from each value of the low frequency region to obtain an updated low frequency region; the updated low frequency region and the other regions constitute a second wavelet coefficient distribution;

The coding transmission module is used to transmit the common coefficients; encode the second wavelet coefficient distribution using an embedded zerotree wavelet coding algorithm, and transmit the coding result.

In some embodiments, the common coefficient N is equal to the difference between the minimum value in the low frequency region and the maximum absolute value in other regions.

In some embodiments, for an input image of 2 ^L ×2 ^M , M≥L≥3, K=L-1 or K=Lm, m>1 is set.

The present invention also provides an image compression device, comprising:

Memory for storing computer programs;

A processor is used to implement the image compression method described in any of the above embodiments when running the computer program.

The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the image compression method described in any of the above embodiments is implemented.

The image compression method, device, and readable storage medium provided by the present invention can at least bring the following beneficial effects:

1. The present invention reduces the size of the value in the low-frequency area by subtracting the common coefficient from each value in the low-frequency area, thereby reducing the initial threshold T of the scan, thereby reducing the number of scans of the embedded zerotree wavelet coding algorithm and the number of tables that need to be transmitted, thereby reducing the overall data transmission volume and improving the lossy compression performance of the image.

2. The present invention can reduce the image compression processing time by performing one or more wavelet transforms on the image, thereby improving the image compression efficiency while ensuring the image compression quality.

BRIEF DESCRIPTION OF THE DRAWINGS

The following will explain the preferred implementation mode in a clear and understandable manner with reference to the accompanying drawings, and further explain the above-mentioned characteristics, technical features, advantages and implementation methods of an image compression method and device, and a readable storage medium.

FIG1 is a flow chart of an embodiment of an image compression method of the present invention;

FIG2-4 is a schematic diagram of the distribution of wavelet transform coefficients obtained by performing multi-level wavelet transform on an image;

FIG5 is a schematic diagram of the main scanning process in the EZW algorithm;

FIG6 is a schematic structural diagram of an image compression device according to an embodiment of the present invention;

FIG. 7 is a schematic structural diagram of another embodiment of an image compression device of the present invention.

Description of Figure Numbers:

100. Wavelet transform module, 200. Public coefficient determination module, 300. Low frequency update module, 400. Encoding transmission module, 10. Memory, 20. Computer program, 30. Processor.

DETAILED DESCRIPTION

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings and other implementation methods can be obtained based on these drawings without creative work.

In order to simplify the drawings, only the parts related to the present invention are schematically shown in each figure, and they do not represent the actual structure of the product. In addition, in order to simplify the drawings and facilitate understanding, in some figures, only one of the parts with the same structure or function is schematically drawn or marked. In this article, "one" not only means "only one", but also means "more than one".

The traditional image compression method based on DWT and EZW algorithm includes the following steps:

1. Perform discrete wavelet transform on the input image

Perform K-level wavelet transform on the input image to obtain the wavelet coefficient distribution. K can be determined according to the size of the image. For example, for an 8*8 image, K can be 3; for a 16*16 image, K can be 4.

Taking an 8*8 image as an example, a two-dimensional wavelet transform (i.e., first-level wavelet transform) is performed to form four regions (as shown in Figure 2), where LL1 is a low-frequency region, representing the main feature information of the image; HL1 and LH1 are medium-frequency regions, representing the secondary feature information of the image; and HH1 is a high-frequency region, representing the detailed feature information of the image. The information of the low-frequency region LL1 is further decomposed, that is, a two-dimensional wavelet transform (i.e., second-level wavelet transform) is performed on LL1 alone, and the wavelet coefficient distribution shown in Figure 3 is obtained. The information of the LL2 region is further decomposed, that is, a two-dimensional wavelet transform (i.e., third-level wavelet transform) is performed on it alone, and the wavelet coefficient distribution shown in Figure 4 is obtained. Through multi-level wavelet transform, the frequency decomposition of the image at different levels can be obtained. Among them, LLx (x=1,2,3) is a low-frequency region, and the others are non-low-frequency regions. The values of the low-frequency region are positive, and other regions may have negative values.

2. Use the EZW algorithm to encode and output the wavelet coefficient distribution

The wavelet coefficient distribution is scanned and outputted multiple times using the EZW algorithm, and the scanning sequence of each time is shown in FIG4 .

The EZW algorithm includes multiple scans of the wavelet coefficient distribution, and each scan includes two stages: main scan and auxiliary scan. The main scan scans the wavelet coefficient distribution to generate important coefficients and tree information, and the auxiliary scan scans the important coefficients and encodes the values of the important coefficients to facilitate reconstruction of the important coefficient values during decoding. It includes:

(1) Determine the initial threshold T

The initial threshold of the scan is T, and its formula is as follows:

Among them, max is the wavelet coefficient with the largest absolute value after wavelet transformation, Represents the largest integer not greater than x.

The threshold value used for the i-th scan is T _i =T/2 ^i-1 , i=1, 2, ..., and as the number of scans increases, the scan threshold value gradually decreases until it is equal to 1, so the T value determines the number of scans of the EZW algorithm.

(2) Perform the main scan

The flow chart of the main scan is shown in FIG5 .

The scanning order is from high level to low level. For the three-level wavelet transform, the scanning order is as follows: LL3,

HL3, LH3, HH3, HL2, LH2, HH2, HL1, LH1, HH1.

During the i-th (i=1, 2, ...) scan, the wavelet coefficients are compared with the threshold _Ti in sequence according to the scan order. If the absolute value of the wavelet coefficient is greater than or equal to the threshold, it is an important coefficient; otherwise, it is an unimportant coefficient and is marked with the following output symbols:

Positive importance coefficient P: the current coefficient is positive and its absolute value is greater than or equal to the threshold;

Negative importance coefficient N: the current coefficient is negative and its absolute value is greater than or equal to the threshold;

Zero tree root ZTR: the current coefficient is an unimportant coefficient, and all descendant coefficients are unimportant coefficients;

Isolated zero IZ: The current coefficient is an unimportant coefficient, but at least one descendant coefficient is an important coefficient.

At the same time, a sub-table is output, in which important coefficients are recorded. At the end of the ith main scan, the important coefficients in the wavelet coefficient distribution are set to zero to avoid encoding them in the next scan.

(3) Perform auxiliary scanning

Scan the sub-table, if the absolute value of the important coefficient is in the interval [T _i , T _i +T _i /2], it is encoded as 0, if the absolute value of the important coefficient is in the interval [T _i +T _i /2, 2T _i ], it is encoded as 1.

The threshold _Ti used in this scan, the obtained tag sequence, and the coded sequence are output for use in decompression.

Then the threshold _Ti is replaced by _Ti /2, and if _Ti /2 is greater than or equal to 1, the next scan is performed, otherwise the process ends. All scan outputs constitute the encoding result of the EZW algorithm.

It can be seen that after the 8*8 image is transformed by the three-level wavelet transform, the main low-frequency information is concentrated in the upper left corner. The size of the initial threshold T of the EZW algorithm is basically determined by the result of the three-level transform. The larger the T value, the more tables need to be transmitted, so the amount of data transmitted is larger. Take the transmission of one more scan table as an example for calculation. Even if the upper left corner is scanned as an important coefficient and the others are not scanned, it needs to transmit an important coefficient and three zero tree roots. Each symbol requires 2 bits to represent, so at least 8 bits of data need to be transmitted.

The present invention proposes a method for reducing an initial threshold value T. By reducing the initial threshold value T, the scanning times of the EZW algorithm are reduced, thereby reducing the number of tables that need to be transmitted, thereby reducing the overall data transmission volume, and improving the image compression effect.

This is explained in detail below.

One embodiment of the present invention, as shown in FIG1 , is an image compression method, comprising:

Step S100 performs K-level wavelet transform on the input image to obtain a first wavelet coefficient distribution;

Step S200 determines the common coefficient N of the low-frequency region of the first wavelet coefficient distribution, where the common coefficient N does not exceed the difference between the minimum value of the low-frequency region and the maximum absolute value of other regions, where the other regions are all non-low-frequency regions in the first wavelet coefficient distribution;

Step S300: subtract the common coefficient N from each value of the low-frequency region to obtain an updated low-frequency region; the updated low-frequency region and other regions constitute a second wavelet coefficient distribution;

Step S400: transmitting the common coefficient N;

Step S500 encodes the second wavelet coefficient distribution using an embedded zerotree wavelet coding algorithm and transmits the coding result.

Specifically, a traditional discrete wavelet transform algorithm is used to perform a K-level wavelet transform on the input image, and the decomposed wavelet coefficients are quantized, and the quantized wavelet coefficients constitute a first wavelet coefficient distribution. K is a positive integer.

Taking an 8*8 image as an example, a secondary wavelet transform (i.e., K=2) is performed, and the distribution of the first wavelet coefficients is shown in Figure 3. Among them, LL2 is the low-frequency area of the first wavelet coefficient distribution. The low-frequency area contains 4 points, and the minimum value of the low-frequency area is the minimum value of these 4 points. LH2, HL2, HH2, HL1, LH1, and HH1 constitute other areas. Assuming that the values of the 4 points in the low-frequency area are 220, 200, 190, and 180 from high to low, and the maximum absolute value of the wavelet coefficients in other areas is 90, then the minimum value of the low-frequency area = 180, and the common coefficient N can be selected as a positive number not exceeding 90 (=180-90), such as N = 90, or 80, etc.

Assuming that N is 90, the updated values of the low-frequency area are 130, 110, 100, and 90 respectively.

If the first wavelet coefficient distribution is subjected to an EZW scan, then according to the aforementioned calculation formula, the initial threshold T is 220. Since the value of the low frequency region is reduced by step S300, the initial threshold T is reduced to 130 when the second wavelet coefficient distribution is subjected to an EZW scan.

The initial threshold T is greatly reduced, which reduces the number of scans of the EZW algorithm, thereby reducing the number of tables that need to be transmitted. In order to correctly restore the first wavelet coefficient distribution on the decoding side, in addition to transmitting the output results of the EZW algorithm scanning the second wavelet coefficient distribution, it is also necessary to increase the transmission of the public coefficient N. Since the amount of data transmitted is reduced more, the overall data transmission amount is reduced, and the image compression effect is improved.

During decoding, the second wavelet coefficient distribution is first restored according to the traditional method, and then each value in the low-frequency region of the second wavelet coefficient distribution is compensated with the received common coefficient N (ie, the common coefficient N is added), thus obtaining the first wavelet coefficient distribution.

N can also take other values. For example, if N is 80, the updated values of the low-frequency area are 140, 120, 110, and 100. When the second wavelet coefficient distribution is subjected to EZW scanning, the initial threshold T is reduced to 140. Compared with 220, the initial threshold T is also reduced a lot, but the reduction is less than when N is 90.

The reason why the common coefficient N is limited to not more than the difference between the minimum value of the low-frequency region and the maximum absolute value of other regions is to ensure that the subsequent update of the low-frequency region value does not affect the zero tree structure of the wavelet coefficient distribution, that is, the zero tree structure of the second wavelet coefficient distribution must be consistent with the zero tree structure of the first wavelet coefficient distribution. Specifically, it is to ensure that the value of the parent node in the wavelet coefficient distribution is not less than the value of the child node. This ensures that the quality of image compression remains unchanged.

By subtracting the common coefficient N from each value in the low-frequency region, the size of the wavelet coefficients in the low-frequency region can be reduced without affecting the zero-tree structure, thereby reducing the initial scanning threshold T and reducing the amount of data transmission.

This embodiment determines a common coefficient and subtracts the common coefficient from each value in the low-frequency area, thereby reducing the initial threshold T of the scan, thereby reducing the number of scans of the embedded zerotree wavelet coding algorithm and the number of tables that need to be transmitted, thereby reducing the overall data transmission volume and improving the lossy compression performance of the image.

In one embodiment, the common coefficient N is equal to the difference obtained by subtracting the maximum absolute value of other regions from the minimum value of the low frequency region.

This allows the initial threshold T of the EZW scan to be minimized without affecting the zerotree structure.

In one embodiment, for an input image of 2 ^L ×2 ^M , M ≥ L ≥ 3, K = L - 1 is set.

For an input image of 2 ^L × 2 ^M , M≥L≥3, a maximum of L levels of wavelet transform can be performed, and K can be set equal to L, and the aforementioned embodiment scheme can be implemented. However, by setting K=L-1 and performing K levels of wavelet transform, one wavelet transform can be reduced, further reducing the image compression processing time and improving the image compression efficiency.

In one embodiment, for an input image of 2 ^L ×2 ^M , M ≥ L ≥ 3, K = Lm, m>1.

For an input image of 2 ^L × 2 ^M , M ≥ L ≥ 3, a maximum of L levels of wavelet transform can be performed. Assuming that L is relatively large, K = Lm can be set to perform K-level wavelet transform, which can reduce m wavelet transforms and the number of embedded zerotree scans, speed up the scanning speed, further reduce the image compression processing time, and improve the image compression efficiency.

This is mainly for relatively large images, which can reduce the number of wavelet transforms. For example, for a 64*64 image, L=M=6, and K can be 4 or 5, which not only improves the image compression efficiency but also ensures the image compression quality. m can be set based on experience.

According to an embodiment of the present invention, as shown in FIG6 , an image compression device includes:

The wavelet transform module 100 is used to perform a K-level wavelet transform on the input image to obtain a first wavelet coefficient distribution;

A common coefficient determination module 200 is used to determine a common coefficient N of a low-frequency region of the first wavelet coefficient distribution, wherein the common coefficient N does not exceed a difference obtained by subtracting a minimum value of the low-frequency region from a maximum absolute value of other regions, where the other regions are all non-low-frequency regions in the first wavelet coefficient distribution;

The low frequency updating module 300 is used to subtract the common coefficient N from each value of the low frequency region to obtain an updated low frequency region; the updated low frequency region and other regions constitute a second wavelet coefficient distribution;

The coding and transmission module 400 is used to transmit the common coefficients; encode the second wavelet coefficient distribution using an embedded zerotree wavelet coding algorithm, and transmit the coding result.

In one embodiment, the common coefficient N is equal to the difference obtained by subtracting the maximum absolute value of other regions from the minimum value of the low frequency region.

In one embodiment, for an input image of 2 ^L × 2 ^M , M ≥ L ≥ 3, set K = L - 1

In one embodiment, for an input image of size 2 ^L ×2 ^M , M ≥ L ≥ 3, K = Lm, m>1.

In one embodiment, during decoding, after the second wavelet coefficient distribution is restored, each value in the low-frequency region of the second wavelet coefficient distribution is compensated accordingly according to the received common coefficients.

It should be noted that the embodiment of the image compression device provided by the present invention and the embodiment of the image compression method provided above are based on the same inventive concept and can achieve the same technical effect. Therefore, other specific contents of the embodiment of the image compression device can refer to the contents of the embodiment of the image compression method provided above.

According to an embodiment of the present invention, as shown in FIG7 , an image compression device includes:

A memory 10 for storing a computer program 20;

The processor 30 is used to implement the image compression method described in any of the above embodiments when running the computer program 20.

The memory 10 may be any internal storage unit and/or external storage device capable of storing data and programs, for example, a plug-in hard disk, a smart memory card (SMC), a secure digital (SD) card, or a flash memory card.

As needed, the processor 10 can be a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a general-purpose processor or other logic devices, etc.

One embodiment of the present invention is a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the image compression method described in the aforementioned embodiment. That is, when part or all of the technical solutions that contribute to the prior art in the aforementioned embodiment of the present invention are embodied in the form of a computer software product, the aforementioned computer software product is stored in a computer-readable storage medium. The computer-readable storage medium can be any physical device or equipment that can carry computer program code. For example, a USB flash drive, a mobile disk, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, etc.

It should be noted that the above embodiments can be freely combined as needed. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.

Claims (10)

1. An image compression method, comprising:

Performing K-level wavelet transformation on an input image to obtain first wavelet coefficient distribution;

Determining a common coefficient N of a low-frequency region of the first wavelet coefficient distribution, wherein the common coefficient N is not more than a difference value obtained by subtracting the maximum value of absolute values of other regions from the minimum value of the low-frequency region, and the other regions are all non-low-frequency regions in the first wavelet coefficient distribution;

Subtracting the common coefficient N from each value of the low frequency region to obtain an updated low frequency region; the updated low frequency region and the other regions form a second wavelet coefficient distribution;

Transmitting the common coefficients;

and adopting an embedded zero tree wavelet coding algorithm to code the second wavelet coefficient distribution, and transmitting a coding result.

2. The image compression method according to claim 1, wherein,

The common coefficient N is equal to a difference obtained by subtracting the maximum value of the absolute values of the other regions from the minimum value of the low frequency region.

3. The image compression method according to claim 1, wherein for an input image of 2 ^L×2^M, M is equal to or greater than L is equal to or greater than 3, and k=l-1 is set.

4. The image compression method according to claim 1, wherein for an input image of 2 ^L×2^M, M is equal to or greater than L is equal to or greater than 3, k=l-M, M >1 is set.

5. The image compression method according to claim 1, comprising:

and in decoding, after recovering the second wavelet coefficient distribution, correspondingly compensating each value of a low-frequency region of the second wavelet coefficient distribution according to the received common coefficient.

6. An image compression apparatus, comprising:

the wavelet transformation module is used for carrying out K-level wavelet transformation on the input image to obtain first wavelet coefficient distribution;

a common coefficient determining module, configured to determine a common coefficient N of a low frequency region of the first wavelet coefficient distribution, where the common coefficient N is not greater than a difference obtained by subtracting a maximum value of absolute values of other regions from a minimum value of the low frequency region, and the other regions are all non-low frequency regions in the first wavelet coefficient distribution;

the low-frequency updating module is used for subtracting the common coefficient N from each value of the low-frequency area to obtain an updated low-frequency area; the updated low frequency region and the other regions form a second wavelet coefficient distribution;

the code transmission module is used for transmitting the common coefficients; and adopting an embedded zero tree wavelet coding algorithm to code the second wavelet coefficient distribution, and transmitting a coding result.

7. The image compression apparatus according to claim 6, wherein,

The common coefficient N is equal to a difference obtained by subtracting the maximum value of the absolute values of the other regions from the minimum value of the low frequency region.

8. The image compression apparatus according to claim 6, wherein,

For an input image of 2 ^L×2^M, M is greater than or equal to L is greater than or equal to 3, K=L-1, or K=L-M, M >1 is set.

9. An image compression apparatus, comprising:

A memory for storing a computer program;

A processor for implementing the image compression method according to any one of claims 1 to 5 when running the computer program.

10. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the image compression method according to any one of claims 1 to 5.