CN104506867A - Sample adaptive offset parameter estimation method and device - Google Patents

Abstract

The present invention discloses a parameter estimation method and device for sampling point self-adaptive offset. The method includes: GPU uses a preset number of threads to first determine the SAO statistical information of all coding blocks in the current coding frame; according to the coding block Dependency relationship, and decision processing is performed to obtain the SAO parameters of all coding blocks of the coding frame. Through the present invention, the GPU is used to make corresponding statistics on the pixel values of the coding blocks to complete the decision-making of the SAO parameters of each coding block, and the acceleration of the processing module is realized by means of the parallel processing capability of the GPU, helping to realize CPU+GPU hybrid coding Acceleration of coding of schemas.

Description

Sampling point adaptive offset parameter estimation method and device

technical field

The present invention relates to the technical field of video compression coding, in particular to a parameter estimation method and device for Sample Adaptive Offset (SAO for short).

Background technique

High Efficiency Video Coding (HEVC) is a new-generation video coding standard jointly proposed by the International Organization for Standardization (ISO) and the ITU Telecommunication Standardization Sector (ITU-T). , this standard can achieve higher video image compression efficiency than previous coding standards. HEVC supports three grades of Main, Main10 and still images, and supports signal encoding in YUV4:2:0 sampling format, which doubles the compression efficiency compared with H.264/AVC. However, due to the high complexity of the HEVC algorithm and the low encoding rate, it is difficult to achieve real-time encoding on a personal computer, so many researchers have begun to use Graphic Processing Unit (GPU) to optimize the encoder , so as to achieve high-speed encoding and achieve better results. For example, the GPU optimization of the main motion estimation module for HEVC, and the GPU acceleration algorithm of the deblocking filter.

Sample Adaptive Offset (SAO for short) is a new filtering algorithm in HEVC. Its algorithm complexity is not high, and it takes up less than 1% of the coding time of the reference software. Currently there is no publicly available GPU acceleration algorithm. However, due to the particularity of its processing, it is difficult to optimize the algorithm through single instruction multiple data instructions. Therefore, during the optimization of the encoder, it is necessary to also perform GPU optimization on SAO. SAO detects the boundary direction and pixel value of the reconstructed frame of the encoded image, and applies different compensation values to points in different directions and pixel intervals, so that the reconstructed frame image is closer to the image to be encoded, thus bringing better results. performance gain.

In the encoding process of HEVC, the image will be divided into encoding blocks of constant size, for example, its size can be configured as 32x32. In the processing of SAO, for each coding block, an optimal decision category and its corresponding four compensation values will be assigned. The types of SAO are divided into two categories: edge offset (Edge Offset, EO for short) and band offset (Band Offset, BO for short). In EO, according to the direction of the boundary, it is divided into four categories: 0 degree, 45 degree, 90 degree and 135 degree. The band in BO is a strip, which is a mapping of pixel values. For example, in an 8-bit image, the pixel value ranges from 0 to 255. The band number of any pixel value x is Due to the concentration of the color, in band compensation, usually only 4 consecutive bands are taken, so an additional value is required to record the starting band number. In addition, for each band, the encoder will The value that needs to be compensated is recorded and encoded into the code stream. An example of SAO parameters for an encoding block is shown in Table 1. The reference mode is set to not refer to the left coding block and the upper coding block to encode the following information, and only when SAO is turned on, the SAO type and the corresponding compensation value need to be coded.

Table 1 SAO parameter format of a coding block

For the above-mentioned problem of relatively low efficiency of the SAO parameter estimation method in the related art, no effective solution has been proposed yet.

Contents of the invention

Aiming at the problem of relatively low efficiency of the SAO parameter estimation method in the related art, the present invention provides a SAO parameter estimation method and device to at least solve the problem.

On the one hand, a method for estimating sampling point adaptive offset SAO parameters is provided, including: using a preset number of threads on a graphics processor GPU to determine the SAO parameters of some coding blocks in the current coding frame; according to the dependence of the coding blocks relationship, and perform decision processing to obtain the SAO parameters of all coding blocks in the coding frame.

Preferably, the GPU uses one thread group warp, and determining the SAO parameters of some coding blocks in the current coding frame includes: each thread in one thread group warp in the GPU corresponds each thread to the The value of the element at the preset position of the coded frame is copied to the cache, wherein the pixel value copied by each coded block in the coded frame is an integer multiple of the warp size of the thread group; in the cache, the Each thread determines the difference between the boundary type of a column of pixel values corresponding to each thread and the pixel value of the corresponding original image frame; according to the column excluded according to the preset rules, the information of the thread corresponding to the column is set is zero; determine boundary statistical EO information and strip statistical BO information according to the statistical information of a thread group warp.

Preferably, the EO information includes: 16 sets of N and E values of a color component, and the BO information includes: 32 sets of N and E values.

Preferably, the points skipped by different color components include:

When the SAO parameter type is EO 0 degree direction, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3;

When the SAO parameter type is EO 90-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3/2;

When the SAO parameter type is EO 135 degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is EO 45-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is BO, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2/1, and the number of columns skipped at the right is 2/1 The number of columns is 3/2;

The number after the slash is the number of points skipped when using the pixel value before deblocking filtering.

Preferably, according to the dependencies of the coding blocks, decision processing is performed to obtain the SAO parameters of all coding blocks of the coding frame including:

The decision-making process is performed in a peak-staggered parallel manner.

On the other hand, a device for estimating sampling point adaptive offset SAO parameters is also provided, which is applied to GPU, including: a determination module, configured to use a preset number of threads to determine the SAO parameters of some coding blocks in the current coding frame; The processing module is configured to perform decision processing according to the dependencies of the coding blocks to obtain the SAO parameters of all coding blocks of the coding frame.

Preferably, the determination module includes:

The first determination unit is used to copy the value of the element at the preset position of each thread corresponding to the encoding frame to the cache when the GPU uses one thread group warp, wherein in the encoding frame The pixel point value copied by each coding block is an integer multiple of the warp size of the thread group;

The second determination unit is used for determining, in the cache, each thread the difference between the boundary type of a column of pixel values corresponding to each thread and the corresponding pixel value of the original image frame;

The setting unit is used to set the column excluded according to the preset rule, and set the information of the thread corresponding to the column to zero;

The third determination unit is configured to determine boundary statistics EO information and stripe statistics BO information according to the statistics information of the one thread group warp.

Preferably, the EO information includes: 16 sets of N and E values of a color component, and the BO information includes: 32 sets of N and E values.

Preferably, the points skipped by the setting module in different color components include:

When the SAO parameter type is EO 0 degree direction, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3;

When the SAO parameter type is EO 90-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3/2;

When the SAO parameter type is EO 135 degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is EO 45-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is BO, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2/1, and the number of columns skipped at the right is 2/1 The number of columns is 3/2;

The number after the slash is the number of points skipped when using the pixel value before deblocking filtering.

Preferably, the processing module performs the decision-making processing in a peak-staggered parallel manner.

Through the present invention, the GPU is used to make corresponding statistics on the pixel values of the coding blocks to complete the decision-making of the SAO parameters of each coding block, and realize the acceleration of the processing module by means of the parallel processing capability of the GPU, helping to realize CPU+GPU hybrid coding Acceleration of coding of schemas.

Description of drawings

The accompanying drawings described here are used to provide a further understanding of the present invention and constitute a part of the application. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations to the present invention. In the attached picture:

Fig. 1 is the flowchart of the SAO parameter estimation method according to the embodiment of the present invention;

Fig. 2 is a schematic diagram of the SAO decision sequence of different blocks according to an embodiment of the present invention;

Fig. 3 is a flow chart of the SAO parameter decision process of a coding block according to an embodiment of the present invention;

Fig. 4 is a structural block diagram of an SAO parameter estimation device according to an embodiment of the present invention.

Detailed ways

Hereinafter, the present invention will be described in detail with reference to the drawings and examples. It should be noted that, in the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

Fig. 1 is the flow chart of the SAO parameter estimation method according to the embodiment of the present invention, as shown in Fig. 1, the method comprises the following steps:

In step S102, the GPU uses a preset number of threads to determine SAO statistical information of all coding blocks in the current coding frame.

In step S104, according to the dependency relationship of the coding block, decision processing is performed to obtain the SAO parameters of all coding blocks in the current coding frame.

In related technologies, the estimation of SAO parameters adopts a serial pattern traversal method, and its efficiency is relatively low. Through the above steps, using GPU for parallel processing can simultaneously maintain the parameter dependencies between coding blocks. Estimate the SAO parameters of multiple coding blocks, and calculate the cost corresponding to multiple compensation value parameter configurations in parallel when estimating the parameters of each coding block, which improves the parallelism of the algorithm and significantly reduces the time complexity of SAO parameter estimation .

Preferably, the GPU can use one thread group warp, and the above step S102 can be performed in the following manner:

Each thread in one thread group warp in the GPU copies the value of each thread corresponding to the element at the preset position of the encoding frame to the cache, wherein each encoding block copy in the encoding frame The pixel value of is an integer multiple of the warp size of a thread group;

In the cache, each thread determines the difference between the boundary type of a column of pixel values corresponding to each thread and the corresponding pixel value of the original image frame;

According to the column excluded according to the preset rules, the information of the thread corresponding to the column is set to zero;

According to the statistical information of the one thread group warp, determine the SAO statistical information of all color components.

Preferably, the SAO information of a color component includes EO information and BO information of the color component, wherein the EO information includes: 16 sets of N and E values of the color component, and the BO information includes: 32 sets of the color component N and E values, where N is the number of pixels of a specific category, and E is the sum of the absolute value of the difference between the original frame pixel value of the pixel position of a specific category and the pixel value when the module is executed.

As another preferred embodiment, the number of points skipped by different color components includes:

When the SAO parameter type is EO 0 degree direction, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3;

When the SAO parameter type is EO 90-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3/2;

When the SAO parameter type is EO 135 degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is EO 45-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is BO, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2/1, and the number of columns skipped at the right is 2/1 The number of columns is 3/2;

The number after the slash is the number of points skipped when using the pixel value before deblocking filtering.

As another preferred implementation manner, in step S104, the decision-making process may be performed in a parallel manner with shifted peaks.

The following will be described in combination with preferred embodiments, and the following preferred embodiments combine the above-mentioned embodiments and preferred implementation modes.

Preferred embodiment one

This preferred embodiment provides a method for estimating SAO parameters on a GPU. In this preferred embodiment, the estimation of SAO parameters is carried out from the reconstructed frame before filtering by the given encoder and the pixel values of the original image frame, and the efficient parallel processing characteristics of GPU are used to firstly calculate the statistical information for each coding block in parallel, and then Blocks make parallel decisions based on dependencies, resulting in a multiplied speedup for that module. A detailed description is given below:

In this preferred embodiment, the thread allocation strategy on the GPU may include:

a) In SAO statistical information calculation and SAO parameter decision-making, the number of threads assigned to each encoding block is the size of a thread group warp of GPU.

b) For each encoding block, the number of pixels copied in each row is an integer multiple of the size of a warp, and pixels beyond the block boundary are not additionally copied.

Preferably, in this preferred embodiment, part of the information of the image boundary is discarded to achieve fast information statistics, and the number of pixels that should be skipped for each coding block is shown in Table 2.

Table 2 Number of points skipped by different color components

It should be noted that the number after the slash is the number of points that should be skipped when using the pixel value before deblocking filtering.

Preferred embodiment two

This preferred embodiment provides a fast SAO decision-making method on GPU. In this preferred embodiment, the method may include two steps of SAO statistical information calculation and SAO parameter decision-making. In the calculation of SAO statistical information, the parallel processing capability of the GPU can be fully utilized, and the entire frame of image can be processed at the same time; in the decision-making of SAO parameters, it is limited to the dependencies between blocks, and the processing is carried out in a peak-staggered parallel processing manner.

As a preferred implementation, in the calculation of SAO statistical information, there is no dependency relationship between each coding block, and it can be processed by a single thread group. If the number of thread groups is large enough, all blocks can be processed at the same time.

Preferably, if the SAO statistical information is calculated in advance, the number of blocks that can be processed at the same time is the blocks that have been generated in the current reconstruction frame;

Preferably, when making SAO parameter decision-making, peak-staggered parallel processing technology is used for parallel processing to fully increase the parallelism.

Preferred embodiment two

This preferred embodiment provides a method for determining snapshot relationship between network tables. In this preferred embodiment, the powerful parallel processing capability of the GPU is used to perform fast and parallel SAO information statistics on the coded image based on the block, and then execute fast SAO parameter decision-making under the condition of ensuring the dependency relationship. Doubling speedup with significant loss.

The following description will be given through specific embodiments.

This preferred embodiment includes: SAO statistical information calculation.

Specifically, in the processing of SAO filtering, the non-boundary pixel points in the coding block will be classified and then filtered, and the compensation value of each category depends on the number of points of this type (denoted as N) and the value of points of this type The difference with the original image (denoted as E). Among the four categories of EO, each category is divided into four subcategories, so there are 16 sets of N and E values for a color component. And BO has 32 groups according to the standard, so there are 32 groups of N and E values. Therefore, 48 pairs of N and E values need to be calculated for each color component. In the calculation of boundary statistics and slice statistics, a warp is allocated for each coded block (size NxN), thereby avoiding additional synchronization.

Preferably, the statistics of boundary information are performed according to the following steps:

Step 1: Each thread in the thread group copies the value of the corresponding position element to the shared cache.

Step 2: From the row of SAO filter processing to the last row defined in Table 2, each thread calculates the boundary type corresponding to a column of pixel values and the corresponding difference with the original image.

Step 3: According to the columns that should be excluded according to Table 2, clear the local statistical information of the threads corresponding to the corresponding columns.

Step 4: Accumulate the statistical information of the thread group, and obtain the N and E values of the four subcategories, that is, the statistical information.

Preferably, in the BO statistical information calculation, the thread group is divided into four groups according to the thread numbers, and each group completes the information statistics of a part of the pixel values. Table 2 shows the number of boundary points of coding blocks that need to be skipped for different color components and SAO types.

This preferred embodiment includes: SAO parameter decision-making.

Preferably, similar to other coding elements, the SAO parameters of a coding block can also refer to the parameters of the block above or to the left, which brings about the dependency between coding blocks, without destroying the dependency Next, the SAO parameter estimation of all coding blocks can be completed in the traversal order as shown in the figure, where the blocks marked with the same value can be processed at the same time, and the block with a larger value needs to wait for the block with a smaller value to be processed.

Under this framework, a thread group is assigned to each decision-making block for decision-making processing. The processing flow is:

Step 1: Initialize, reset the context model of all rows, and set the parameter array of the reference block to be empty, and set the sum of the horizontal and vertical coordinates of the position of the next processing block to seq=0.

Step 2: For all blocks whose sum of horizontal and vertical coordinates is seq, record c as the state of the context model, and perform SAO parameter estimation according to the process shown in the figure;

Step 3: Write the SAO parameters of all the blocks of the current decision into the global storage, and use the shared cache backup for future reference;

Step 4: If all blocks are processed, end, otherwise seq increases by 1 and continues to cycle from step 2.

It should be noted that the steps shown in the flowcharts of the accompanying drawings may be performed in a computer system, such as a set of computer-executable instructions, and that although a logical order is shown in the flowcharts, in some cases, The steps shown or described may be performed in an order different than here.

As used below, the terms "sub-module" and "module" may be a combination of software and/or hardware that realizes a predetermined function. Although the systems and methods described in the following embodiments are preferably implemented in software, implementations in hardware, or a combination of software and hardware are also possible and contemplated.

In another embodiment, a software for estimating SAO parameters is also provided, and the software is used to implement the technical solutions described in the above embodiments and preferred embodiments.

In another embodiment, a storage medium is also provided, in which the above-mentioned SAO parameter estimation software is stored, and the storage medium includes but is not limited to: optical disks, floppy disks, hard disks, rewritable memories, and the like.

The embodiment of the present invention also provides an SAO parameter estimation device, which can be used to implement the above SAO parameter estimation method and preferred implementation modes, which have already been explained and will not be described in detail, and the SAO parameter estimation device is described below The modules involved in are described. As used below, the term "module" may be a combination of software and/or hardware that realizes a predetermined function. Although the systems and methods described in the following embodiments are preferably implemented in software, implementations in hardware, or a combination of software and hardware are also possible and contemplated.

Fig. 4 is a structural block diagram of an SAO parameter estimation device according to an embodiment of the present invention, the device can be applied to GPU, as shown in Fig. 4, the device includes: a determination module 42, a processing module 44, the above structure will be described in detail below .

The determination module 42 is used to determine the SAO statistical information of all coding blocks in the current coding frame by using the preset number of threads; the processing module 44 is connected to the determination module 42 and is used to perform decision processing according to the dependencies of the coding blocks , to obtain the SAO parameters of all coded blocks of the coded frame.

Preferably, the determination module may include:

The first determination unit is used for copying the value of the element at the preset position of each thread corresponding to the encoding frame to the cache when the GPU uses one thread group warp, wherein in the encoding frame The pixel point value copied by each coding block is an integer multiple of the warp size of the thread group;

The second determination unit is used for determining, in the cache, each thread the difference between the boundary type of a column of pixel values corresponding to each thread and the corresponding pixel value of the original image frame;

The setting unit is used to set the column excluded according to the preset rule, and set the information of the thread corresponding to the column to zero;

The third determination unit is configured to determine boundary statistics EO information and stripe statistics BO information according to the statistics information of the one thread group warp.

Preferably, the EO information includes: 16 sets of N and E values of a color component, and the BO information includes: 32 sets of N and E values.

Preferably, the points skipped by the setting module in different color components include:

When the SAO parameter type is EO 0 degree direction, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3;

When the SAO parameter type is EO 90-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2, and the column skipped at the right The number is 3/2;

When the SAO parameter type is EO 135 degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is EO 45-degree direction, the number of rows skipped at the bottom in the luma component is 4, and the number of columns skipped at the right is 5; the number of rows skipped at the bottom in the chroma component is 2, and the number of columns skipped at the right is 3;

When the SAO parameter type is BO, the number of rows skipped at the bottom in the luma component is 4/3, and the number of columns skipped at the right is 5/4; the number of rows skipped at the bottom in the chroma component is 2/1, and the number of columns skipped at the right is 2/1 The number of columns is 3/2;

The number after the slash is the number of points skipped when using the pixel value before deblocking filtering.

Preferably, the processing module performs the decision-making processing in a peak-staggered parallel manner.

Through the above embodiments, a method and device for estimating SAO parameters, and a method for estimating SAO parameters on a GPU are provided. Its processing process is divided into two parts: SAO information calculation and SAO parameter decision-making. According to the processing characteristics of the GPU, properly arranging the workload of the threads on the GPU and the algorithm processing flow can achieve a higher speed-up ratio. In the calculation of SAO information, the computing power of the GPU is fully utilized to process multiple blocks at the same time, and in the processing of a block, an additional copy of a single pixel value is avoided, thereby achieving a relatively small loss. High speedup, meanwhile, proposes boundary point information that should be skipped before deblocking filtering. In SAO decision-making, the dependence of the context coding model between rows is eliminated, and a higher degree of parallelism is achieved by using peak-staggered parallel processing.

It should be noted that these technical effects are not possessed by all the above-mentioned implementation manners, and some technical effects can only be obtained by certain preferred implementation manners.

Obviously, those skilled in the art should understand that each module or each step of the above-mentioned present invention can be realized by a general-purpose computing device, and they can be concentrated on a single computing device, or distributed in a network formed by multiple computing devices Optionally, they can be implemented with program codes executable by computing devices, so that they can be stored in storage devices and executed by computing devices, or they can be made into individual integrated circuit modules, or their Multiple modules or steps are implemented as a single integrated circuit module. As such, the present invention is not limited to any specific combination of hardware and software.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. a sampled point self adaptation skew SAO method for parameter estimation, is characterized in that, comprising:

Graphic process unit GPU uses and presets number of threads, determines the SAO parameter of first encoding block in current encoded frame;

According to the dependence of described encoding block, carry out decision-making treatment, obtain the SAO parameter of all encoding blocks of described coded frame.

2. method according to claim 1, is characterized in that, described GPU uses 1 sets of threads warp, determines that the SAO parameter of the code segment block in current encoded frame comprises:

The numerical value that this each thread is corresponded to the element of the predeterminated position of described coded frame by each thread in 1 sets of threads warp in described GPU copies buffer memory to, wherein, in described coded frame, the pixel numerical value of each encoding block copy is the integral multiple of a described sets of threads warp size;

In described buffer memory, described each thread determines the difference of the boundary types of the row pixel value that this each thread is corresponding and the pixel value of corresponding described original image frame;

According to the row that preset rules is got rid of, the information of thread corresponding for described row is set to zero;

According to the statistical information of a described sets of threads warp, determine Boundary Statistic EO information and band statistics BO information.

3. method according to claim 2, is characterized in that, described EO information comprises: 16 groups of N and E values of a color component, and described BO information comprises: 32 groups of N and E values.

4. method according to claim 3, is characterized in that, counting that different colours component is skipped comprises:

When SAO parameter type is EO 0 degree of direction, in luminance component, line number is skipped in bottom is 4/3, and it is 5 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3 that columns is skipped on the right;

When SAO parameter type is EO 90 degree of directions, in luminance component, line number is skipped in bottom is 4, and it is 5/4 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3/2 that columns is skipped on the right;

When SAO parameter type is EO 135 degree of directions, in luminance component, line number is skipped in bottom is 4, and it is 5 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3 that columns is skipped on the right;

When SAO parameter type is EO 45 degree of directions, in luminance component, line number is skipped in bottom is 4, and it is 5 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3 that columns is skipped on the right;

When SAO parameter type is BO, in luminance component, line number is skipped in bottom is 4/3, and it is 5/4 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2/1, and it is 3/2 that columns is skipped on the right;

That skips when wherein slash numeral is below and uses pixel numerical value before block elimination filtering counts.

5. method according to any one of claim 1 to 4, is characterized in that, according to the dependence of described encoding block, carries out decision-making treatment, and the SAO parameter obtaining all encoding blocks of described coded frame comprises:

According to avoiding the peak hour, parallel mode carries out described decision-making treatment.

6. a sampled point self adaptation skew SAO parameter estimation apparatus, is applied to image processor GPU, it is characterized in that, comprising:

Determination module, for using default number of threads, determines the SAO parameter of the code segment block in current encoded frame;

Processing module, for the dependence according to described encoding block, carries out decision-making treatment, obtains the SAO parameter of all encoding blocks of described coded frame.

7. device according to claim 6, is characterized in that, described determination module comprises:

First determining unit, when using 1 sets of threads warp for described GPU, numerical value for the element this each thread being corresponded to the predeterminated position of described coded frame copies buffer memory to, wherein, in described coded frame, the pixel numerical value of each encoding block copy is the integral multiple of a described sets of threads warp size;

Second determining unit, in described buffer memory, described each thread determines the difference of the boundary types of the row pixel value that this each thread is corresponding and the pixel value of corresponding described original image frame;

Setting unit, for the row got rid of according to preset rules, is set to zero by the information of thread corresponding for described row;

3rd determining unit, for the statistical information according to a described sets of threads warp, determines Boundary Statistic EO information and band statistics BO information.

8. device according to claim 7, is characterized in that, described EO information comprises: 16 groups of N and E values of a color component, and described BO information comprises: 32 groups of N and E values.

9. method according to claim 8, is characterized in that, describedly arranges counting that module skips when different colours component and comprises:

When SAO parameter type is EO 0 degree of direction, in luminance component, line number is skipped in bottom is 4/3, and it is 5 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3 that columns is skipped on the right;

When SAO parameter type is EO 90 degree of directions, in luminance component, line number is skipped in bottom is 4, and it is 5/4 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3/2 that columns is skipped on the right;

When SAO parameter type is EO 135 degree of directions, in luminance component, line number is skipped in bottom is 4, and it is 5 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3 that columns is skipped on the right;

When SAO parameter type is EO 45 degree of directions, in luminance component, line number is skipped in bottom is 4, and it is 5 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2, and it is 3 that columns is skipped on the right;

When SAO parameter type is BO, in luminance component, line number is skipped in bottom is 4/3, and it is 5/4 that columns is skipped on the right; In chromatic component, line number is skipped in bottom is 2/1, and it is 3/2 that columns is skipped on the right;

That skips when wherein slash numeral is below and uses pixel numerical value before block elimination filtering counts.

10. the device according to any one of claim 6 to 9, is characterized in that, described processing module carries out described decision-making treatment according to parallel mode of avoiding the peak hour.