CN103220529B - A video codec loop filtering implemented method - Google Patents

A video codec loop filtering implemented method Download PDF

Info

Publication number
CN103220529B
CN103220529B CN201310128415.2A CN201310128415A CN103220529B CN 103220529 B CN103220529 B CN 103220529B CN 201310128415 A CN201310128415 A CN 201310128415A CN 103220529 B CN103220529 B CN 103220529B
Authority
CN
China
Prior art keywords
block
filtering
boundary
filter
pixel
Prior art date
Application number
CN201310128415.2A
Other languages
Chinese (zh)
Other versions
CN103220529A (en
Inventor
马思伟
贺坚强
高文
Original Assignee
北京大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京大学 filed Critical 北京大学
Priority to CN201310128415.2A priority Critical patent/CN103220529B/en
Publication of CN103220529A publication Critical patent/CN103220529A/en
Application granted granted Critical
Publication of CN103220529B publication Critical patent/CN103220529B/en

Links

Abstract

本发明涉及多媒体视频编解码技术领域,尤其涉及一种视频编解码端的环路滤波的方法,所述方法包括利用编解码端块划分信息得到图像块划分结构,环路滤波则在块边界进行;若边界包含编码块、预测块和变换块边界,则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策,最后执行像素级滤波,滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度,本发明的技术效果在于:本方案能有效提高视频编码的主客观质量,在AVS2.0实施例参考软件RD2.0-ADI上,本方案和装置能够提高0.9%的编码效率,同时能减少编解码端滤波边界数和复杂度。 The present invention relates to a multimedia video codec technology, and in particular relates to a method for video encoding and decoding side loop filtering, the method comprising using a codec end block division information to obtain an image block division structure, loop filtering at block boundaries; If the boundaries of the coding block comprises a prediction block and a transform block boundary, the utilization mode, motion information, quantization parameter, and a residual other boundary decision stage filter, and finally perform pixel-level filtering, the filter may be a direction vertical to the boundary pixel according to the local characteristics may also be and a block mode or boundary at other angles, the technical effect of the present invention is that: the present embodiment can improve the subjective and objective quality of the video encoding, with reference to embodiments in software RD2.0-ADI AVS2.0, and apparatus of the present embodiment can be improved 0.9 % of coding efficiency, while reducing the number of boundary codec end of the filter and complexity.

Description

一种视频编解码环路滤波的实现方法 A video codec loop filtering implemented method

技术领域 FIELD

[0001] 本发明涉及多媒体视频编解码技术领域,尤其涉及一种视频编解码端的环路滤波的方法。 [0001] The present invention relates to a multimedia video codec technology, and in particular relates to a loop filtering method for a video codec end.

背景技术 Background technique

[0002] 在当前基于块的混合编码框架中,视频编码以每个块的率失真优化(RDO)决策选择该块的最优模式从而得到当前编码帧编码性能最优的近似。 [0002] In this hybrid coding frame block-based video coding of each block at a rate distortion optimization (RDO) decision to select the optimal mode of the block to obtain a current frame coded optimal coding performance approximation. 但是视频编码中因为考虑了空域和时域的相关性,这样当前块的最优不能保证整个序列编码性能的最优,一些编码工具如环路滤波可以改善这种局部的最优给全局带来的主客观质量的影响。 However, considering that the video coding of the spatial and temporal correlation, most of the current block can not be guaranteed such optimal performance of the entire coding sequence, some coding tools such as the local loop filtering may improve the global optimum bring the influence of subjective and objective quality. 通过改善当前帧在编码时因为分块带来的块效应,减少了主观上的振铃效应和块效应,同时滤波后的帧将作为后续编码帧参考帧,从而提高后续帧的主客观编码质量。 By improving the blocking effect because the current frame coding block, resulting in less ringing and block the subjective effects, while the filtered frame as a reference frame subsequent coding frame, thereby improving the subjective and objective quality of the coding of subsequent frames . 故环路滤波技术在视频编码中有着十分重要的地位。 Therefore, loop filter technology has a very important position in video coding.

[0003] 在现有已发布的视频编码标准中,如H. 264其环路滤波是在4X4块边界、在AVS 中,则是在8X8边界进行的,这是因为在这些标准中,变换块的尺寸是固定的,且不大于编码单元块(宏块)以及预测块的尺寸,所以变换块的边界肯定包含编码单元块以及预测块边界。 [0003] In conventional video coding standards have been published, such as H. 264 loop filtering in which 4X4 block boundary in the AVS, is carried out in the border 8X8, since these standards, transform block the size is fixed, not more than the coding unit block (macro block) and a prediction block size, the boundary of the transform block comprises a positive prediction encoding unit block and the block boundary. 然而在高效视频编码标准和下一代AVS标准AVS2. 0中,编码单元(Coding Unit,⑶) 是视频编码的基本单元。 However, in the next generation of high efficiency video coding standards and AVS standard AVS2. 0, the encoding unit (Coding Unit, ⑶) is the basic unit of video encoding. 编码单元的采用四叉递归划分模式,编码单元最大可达64X64 (如图1所示),而不再是16X16。 Coding uses recursive partitioning unit quad mode, the coding unit up to 64X64 (Figure 1), rather than 16X16. 对每个编码单元中,通过预测、变换量化、熵编码、后处理等编码工具来进行块编码。 For each coding unit by the prediction, transform quantization, entropy encoding, the encoding tool to post-processing block encoded. 其中编码单元在做预测的时候,编码单元可以被划分成不同大小的预测单元(Prediction Unit,PU),预测单元是预测的基本单元。 Wherein the coding unit is predicted to do when the coding unit may be divided into different sizes of the prediction unit (Prediction Unit, PU), the prediction unit is the basic unit of prediction. 编码单元中帧内预测和帧间预测单元的划分分别如图2和图3所示,预测块的尺寸从64X64、64X32、32X64、32X32 - 直到8X8,甚至还包括非对称预测块(AMP)。 And an intra prediction coding unit divided inter prediction unit, respectively 2 and 3, the size of the prediction block from the 64X64,64X32,32X64,32X32 - 8X8 up, even asymmetrical further comprising a prediction block (AMP). 同时变换块的尺寸不再固定为4X4或者8X8,而且和⑶的尺寸相关,这样变换块的尺寸包含从32X32、16X16到8X8。 Meanwhile transform block size is 4X4 or 8X8 longer fixed, and the size and related ⑶, so transform block size comprises from 32X32,16X16 to 8X8. 这样原来的固定在所有的4X4或者8X8边界进行滤波的方式将不再符合环路滤波去块效应的目的。 A fixed manner so that the original filtering all 4X4 or 8X8 no longer meets the object boundary loop filtering deblocking.

发明内容 SUMMARY

[0004] 本发明实施例提供一种环路滤波技术、装置及系统,以减少块效应,提高主客观编码视频质量和编码效率,同时降低编码解码端复杂度。 [0004] An embodiment provides a loop filtering techniques, apparatus and systems of the present invention, in order to reduce the blocking effect, and improve subjective and objective quality of the encoded video coding efficiency, reducing the coding complexity of the decoder.

[0005] 为达到上述目的,本发明提供了一种视频编解码环路滤波的实现方法方案,具体实现步骤包括: [0005] To achieve the above object, the present invention provides a video codec implementation scheme loop filtering, specific implementation steps comprising:

[0006] 所述环路滤波方法包括如下步骤: [0006] The loop filtering method comprising the steps of:

[0007] 步骤1)利用编解码端块划分信息得到图像块(包括编码块、预测块和变换块)划分结构,环路滤波则在块(如编码块、预测块和变换块)边界进行,若编码块的编码残差为0 (即CBP = 0),则仅在编码块和预测块边界进行滤波;滤波流程为预定义滤波分割块,把图像划分为滤波分割单元块(如16X16块大小、也可以是最大编码单元块或图像帧等),可以预定义或在码流中说明滤波分割块的尺寸,为了统一滤波器设计,滤波顺序为依次对每个滤波分割块进行,先垂直边界后水平边界,即先对滤波分割块内的所有垂直边界进行滤波再对水平边界进行滤波;同时预定义最小滤波单元(如8X8块大小),尺度小于最小滤波单元的块边界跳过滤波; [0007] Step 1) using a codec block end image block division information obtained (including a coding block, a transform block and a prediction block) divided structure, loop filtering at block (encoding block, a prediction block and a transform block) boundaries, If the encoded residual coding block is 0 (i.e., CBP = 0), only the filtered prediction block and the encoding block boundary; filtering process to filter a predefined division block, dividing the image into a filtering unit block (e.g., block size 16X16 , the maximum coding unit may be a block or image frames, etc.), or may be predefined dimension filtered sub-blocks in the bitstream, for the unified filter design, the filter order of sequentially dividing each filtered block, the first vertical boundary after the horizontal boundary, i.e. first of all the vertical boundary filtering block division boundary are filtered and then the filter level; while predefined minimum filter unit (e.g., 8X8 block size), the minimum dimension of less than the block boundary filtering element skip filtering;

[0008] 步骤2)若边界包含编码块、预测块和变换块边界,则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策, [0008] Step 2) If the encoding block boundary, a prediction block and a transform block boundary, the utilization mode, motion information, quantization parameter, and a residual other boundary decision stage filter,

[0009] 2. 1、若决策需要滤波该边界,则进行接下来的像素级滤波过程; [0009] 2.1, if the decision boundary filtering required, then perform the next pixel level filtering process;

[0010] 2. 2、否则跳过该边界滤波;对需要进行滤波的块边界处进行像素级滤波决策以决定每个像素行是否需要滤波以及采用的滤波强度(BS); [0010] 2.2, otherwise skip the boundary filtering; pixel-level filtering decision block boundary filtering is required in order to determine whether each pixel row requires filtering and the filtering strength used (the BS);

[0011] 步骤3)最后执行像素级滤波,滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度,其中参与图像滤波的块结构,包括但不限于编码块、预测块和变换块;滤波分割块为方便滤波等操作,而对图像进行统一尺寸的划分方式,如对图像可以按照16X16尺寸滤波分割块划分。 [0011] Step 3) the pixel-level filtering is performed last filtering direction may be perpendicular to the boundary may be at other angles according to the local pixel characteristic or block mode and a boundary, which participate in the block structure of the image filtering, including but not limited to, the coding block, a prediction block and a transform block; filtering divided blocks to facilitate filtering and other operations, while the image division manner a uniform size, such as block division image may be divided in accordance with the filter size 16X16.

[0012] 所述步骤1)还包括:环路滤波发生的边界在块(如编码块、预测块和变换块)边界,当编码块的编码残差为零(即CBP = 0)时,则只在编码块和预测块边界进行滤波,同时定义滤波的最小尺寸如8X8块大小。 [0012] step 1) further comprises: loop filtering occurs in a boundary block (encoding block, a prediction block and a transform block) boundaries, when the encoded residual coded block is zero (i.e., CBP = 0), then only the coding block and a prediction filtering on the block boundary, while the minimum size of the filter is defined as 8X8 block size. 例如对于AVS2实施例,编码块的尺寸包含了从64X64 到8X8大小,根据RDO选择,采用四叉树迭代递归划分,是编码处理的基本单元;预测块为在编码块的基础上的进一步根据预测特性的进一步划分;变换块也是在编码块的基础上根据残差的分布特性的进一步划分。 For example AVS2 embodiment, the size of the coding block contains from 64X64 to 8X8 size, according RDO selection quadtree iterative recursive partitioning, it is the basic unit of encoding processing; prediction block based on a coding block on further the prediction characteristics further divided; transform block is further divided according to the distribution characteristics of the residual block on the basis of the coding.

[0013] 所述步骤1)还包括:对于整帧图像划分为滤波分割单元块(如16X16大小或最大编码单元块),可以预定义或者在码流中说明滤波分割块的尺寸,按照扫描依次对每个滤波分割块进行滤波,先对垂直边界,在水平边界,检测边界是否包含编码块、预测块和变换块边界,若包含,则滤波该边界,否则跳过滤波该边界。 [0013] step 1) further comprises: for filtering the whole frame is divided into the divided unit block (e.g., 16X16 block size or a maximum coding unit), or may be predefined block size division filter described in the code stream, in accordance with the scanning sequence for each filtered for filtering the divided blocks, the first vertical boundary, a horizontal boundary, whether the boundary comprises detecting a coding block, a prediction block and a transform block boundary, if included, the filtering of the boundary, the boundary otherwise skip filtering.

[0014] 所述步骤2)还包括:根据滤波边界两边块的模式、运动信息、量化参数以及残差等来决定当前块边界是否需要滤波,若边界级决策需要滤波该边界,则进行接下来的像素级滤波过程,否则跳过该边界滤波,若块边界两边块P和Q有帧内预测模式,则需要执行滤波;若P、Q块均为P帧的帧间预测块,其残差均为〇,参考帧相同且运动矢量各分量小于一个像素,则决策该块边界不需要滤波,否则均需要滤波。 [0014] step 2) further comprising: a filter according to the pattern on both sides of the boundary block, motion information, quantization parameters, and other residuals to determine whether the current block boundary filtering, if required filtering the boundary level decision boundary, is performed next pixel level filtering process, or the boundary filtering is skipped, if both sides of the block boundary blocks P and Q have the intra prediction mode, the filter needs to be performed; if P, Q blocks are inter-frame prediction block P, the residual are square, the identical reference frame and the motion vector of each component is smaller than a pixel, the filter need not then decision block boundary otherwise require filtering.

[0015] 所述步骤2)还包括:滤波决策需要滤波的块边界每行像素,利用边界两边像素的跳变和块平均QP值得到的索引所对应的阈值表对应值比较,来选择滤波强度;像素级滤波强度取值范围0-4, BS = 0表示不需要滤波,BS = 4,表示较为平坦区域,需要进行较强的平滑滤波。 [0015] step 2) further comprises: a block boundary filtering decisions need filtered pixels per line, using the threshold value table corresponding to the values ​​of the transition and the block average QP boundaries on both sides of the pixel worth to the index corresponding to the comparison, to select the filter strength ; stage filter pixel intensity range 0-4, BS = 0 indicates that no filtering, BS = 4, represents a relatively flat area, the need for a strong smoothing filter.

[0016] 所述步骤3)还包括:像素的位置选择对应的FIR滤波器,滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度;不同滤波强度和不同像素位置所对应的FIR滤波器使用;同时可以利用局部图像的纹理特性和边界两边块的模式信息决定滤波的方式是垂直于边界或者和边界成一定角度 [0016] step 3) further comprising: selecting a pixel position of the FIR filter, the filter corresponding to the direction vertical to the boundary may be at other angles according to the local characteristics of the pixel or block mode and a boundary; different filtering strength and different pixel positions using the corresponding FIR filter; texture features may be utilized simultaneously and on both sides of the block mode boundary information determined partial image filtering method is perpendicular to the boundary or angle the boundary and

[0017] 本发明的技术效果在于:本方案能有效提高视频编码的主客观质量,在AVS2. 0实施例参考软件RD2. O-ADI上,本方案和装置能够提高0. 9%的编码效率,同时能减少编解码端滤波边界数和复杂度。 [0017] A technical effect of the present invention is that: the present embodiment can improve the subjective and objective quality of the video encoding, for example, in embodiments with reference to the software RD2 AVS2 0 on O-ADI, and apparatus of the present embodiment can improve the coding efficiency of 0.9%. , while reducing the number of boundary codec end of the filter and complexity.

附图说明 BRIEF DESCRIPTION

[0018] 当结合附图考虑时,通过参照下面的详细描述,能够更完整更好地理解本发明以及容易得知其中许多伴随的优点,但此处所说明的附图用来提供对本发明的进一步理解, 构成本发明的一部分,本发明的示意性实施例及其说明用于解释本发明,并不构成对本发明的不当限定,如图其中: [0018] when considered in conjunction with the accompanying drawings, by reference to the following detailed description can be more complete and better understanding of the present invention wherein the readily know many advantages attendant, the drawings herein are used to provide further explanation of the present invention. understood that constitute part of the invention, exemplary embodiments of the present invention are used to explain the present invention without unduly limiting the present invention, wherein FIG:

[0019] 图1为AVS2现有技术编码单元划分示意图; [0019] FIG. 1 is a prior art coding unit splitter AVS2 schematic;

[0020] 图2为AVS2现有技术变换块单元划分示意图; [0020] FIG. 2 is a prior art conversion AVS2 schematic block division unit;

[0021 ] 图3为AVS2现有技术预测块单元划分示意图; [0021] FIG. 3 is a prior art prediction AVS2 schematic block division unit;

[0022] 图4为虚线为AVSl采用的,对所有包括8X8变换块边界进行滤波,则本方案仅仅在实线(编码块、预测块和变换块)边界进行滤波; [0022] FIG. 4 is a dotted line AVSl employed, including all 8X8 transform filter block boundaries, the filtering according to the present embodiment only in the solid line (the coding block, a prediction block and a transform block) boundary;

[0023] 图5为滤波边界处参与滤波的像素; [0023] FIG. 5 is a boundary pixel filtering filtered participation;

[0024] 图6为滤波分割块(取16X16尺寸为例)内滤波顺序,先垂直边界,后水平边界; [0024] FIG. 6 is a block division filter (size 16X16 taken as an example) the filtering order, the first vertical boundary, a horizontal boundary post;

[0025] 图7为本发明支持的一种滤波实例流程图; [0025] FIG. 7 is a flowchart of an example of the filter support of the present invention;

[0026] 图8为本发明块边界像素滤波流程图。 [0026] FIG. 8 flowchart block boundary pixel filtering of the present invention.

[0027] 下面结合附图和实施例对本发明进一步说明。 Figures and examples further illustrate the present invention [0027] The following binding.

具体实施方式 Detailed ways

[0028] 显然,本领域技术人员基于本发明的宗旨所做的许多修改和变化属于本发明的保护范围。 [0028] Obviously, many modifications and variations of skill in the art based on the gist of the present invention is made to fall within the scope of the present invention.

[0029] 编码端: [0029] an encoder:

[0030] 如图1至图3所示,本发明实施例AVS2,在编码端,当一帧图像在完成预测、变换量化、和熵编码和模式信息的编码后,再经过反量化反变换后可以得到编码重构帧。 [0030] As shown in FIG. 1 to FIG. 3, the present invention AVS2 embodiment, at the encoding end, after the completion of an image prediction, transform quantization, and entropy coding and coding mode information, and then through the inverse quantization inverse transformed a reconstructed frame can be encoded. 利用编解码端的模式信息和块分割信息块得到当前重构图像帧的划分方式,设置当前帧最小滤波单元,将图像划分为滤波分割块。 End using the codec mode information and the block division information block division manner to obtain a reconstructed image of the current frame, setting a minimum current frame filtering unit filtering the image into the divided blocks. 帧级滤波流程为对每一个滤波分割块,按照图7所示的流程,先对滤波分割块内的所有垂直边界,依次检测其是否为编码块、预测块或变换块边界, 若是则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策。 Frame-level filtering process is a filtering for each divided block, according to the flow shown in FIG. 7, first of all vertical block boundaries within the division filtering sequentially detecting whether the coding block, or prediction block transform block boundary, if the mode using , motion information, quantization parameters and residuals and other boundary level filtering decisions. 若边界级滤波决策需要滤波该垂直边界,则对该垂直边界按照图8所示执行像素级滤波强度决策以及像素级滤波,具体的滤波强度BS决策过程和不同BS下的滤波过程如前所述。 If the decision boundary filtering stage filters the required vertical boundary, the boundary of the vertical filtering stage performing pixel intensity and the pixel decision stage filter shown in Figure 8, the filtering process under specific filter strength BS decision process as described above and different BS . 执行完一个滤波分割块内的所有垂直边界滤波后,再类似的方式执行水平边界滤波。 After performing a dividing filter all vertical boundary filtering in the block boundary filtering and then performs horizontal in a similar manner. 滤波完一个滤波分割块后, 判断是否滤波完一帧,若没有,则按照设定的扫描顺序准备下一个滤波分割块滤波,重复循环直至遍历完当前帧全部滤波分割块。 After filtering the divided block a filter, the filter is determined whether a finished, if not, the scanning order of the set is prepared for the next divided block filter filtering cycle is repeated until all of the current frame is finished filter division block traversal.

[0031] 解码端: [0031] The decoding end:

[0032] 对于解码端,也是依次对每个滤波分割块,经过预测补偿加上反量化反变换后残差得到解码重构块执行类似编码端进行的滤波决策以及滤波过程从而得到解码端重构图像滤波后的图像,用于显示、存储或提供给其他帧参考。 [0032] For the decoding side is sequentially divided blocks each filtered through compensation prediction inverse quantization plus inverse transform residuals obtained after reconstruction block decoder performs filtering and the filtering process is similar to the decision-making for the encoder to obtain the decoding end Reconstruction filtering the image after the image, for display, storage, or to other reference frames.

[0033] 本发明实施例上述方法、装置及系统技术方案具有如下有益效果:本专利提出了一套环路滤波技术。 [0033] Example embodiments of the present invention, the above-described method, apparatus and system aspect has the following advantages: This patent proposes a loop filtering techniques. 在编解码进行环路滤波时,通过利用编解码端块划分信息得到图像块(包括编码块、预测块和变换块)划分结构,环路滤波则在块(如编码块、预测块和变换块) 边界进行,若编码块的编码残差为零(即CBP = 0),则仅在编码块和预测块边界进行滤波; 滤波流程为预定义滤波分割块,把图像划分为滤波分割单元块(如16X16块大小、也可以是最大编码单元块或图像帧等),可以预定义或在码流中说明滤波分割块的尺寸,为了统一滤波器设计,滤波顺序为依次对每个滤波分割块进行,先垂直边界后水平边界,即先对滤波分割块内的所有垂直边界进行滤波再对水平边界进行滤波;同时预定义最小滤波单元,尺度小于最小滤波单元的块边界跳过滤波;若边界包含编码块、预测块和变换块边界,则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策, When the codec performing loop filtering, to obtain an image block (including coding block, a transform block and a prediction block) is divided by the structure dividing information using the codec end block, loop filtering at block (encoding block, a prediction block and a transform block ) boundaries, when the encoded residual coded block is zero (i.e., CBP = 0), only the coding block and a prediction filtering on the block boundary; filtering process to filter a predefined division block, dividing the image into a filtering unit block ( the 16X16 block size, the maximum coding unit may be a block or image frames, etc.), or may be predefined dimension filtered sub-blocks in the bitstream, for the unified filter design, the filter order is sequentially divided each filtered block , the first vertical boundary horizontal boundary, i.e. first of all the vertical boundary filtering block division boundary are filtered and then the filter level; while predefined minimum filter unit, the smallest dimension of less than the block boundary filtering element skip filtering; if boundary comprising coding block, a prediction block and a transform block boundary, the utilization mode, motion information, quantization parameter, and a residual other boundary decision stage filter, 若决策需要滤波该边界,则进行接下来的像素级滤波过程,否则跳过该边界滤波;对需要进行滤波的块边界处阈值(如附表1所示)和像素局部纹理特征进行像素级滤波决策以决定每个像素行是否需要滤波以及采用的滤波强度(BS);最后利用滤波器(附表2或附表3所示)设计执行像素级滤波,滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度,从而能更好的自适应去除块效应和提高编码效率。 If the decision boundary filtering required, then perform the next pixel level filtering process, otherwise skip the boundary filtering; threshold value (as shown in Table 1) at the block boundary filtering and the need for local texture features of the pixels for pixel-level filtering decision to determine whether each pixel row requires filtering and filter strength (BS) used; Finally, a filter (as shown in Table 2 or Table 3) designed to perform pixel-level filtering, the filter may be a direction vertical to the boundary may be partial pixel or block mode and boundary characteristics at other angles, and thus better able to adaptively removing blocking effect and improve encoding efficiency. 本方案在AVS2. O实施例RD2. O上,本方案和装置能够提高0. 9%的编码效率,同时能减少编解码端滤波边界数和复杂度。 The program embodiment AVS2. O RD2. O on, the program, and the coding efficiency can be improved apparatus 0.9%, while the end of the filter can reduce the encoding and decoding complexity and the number of boundaries.

[0034] 实验结果: [0034] The results:

[0035] 本发明实施例在RD2. 0(AVS2标准的参考软件)上完成技术实现。 Example [0035] The present invention has been completed on the technology (AVS2 reference standard software) RD2. 0. 在实验中,采用AVS2 的通用测试条件,RA配置。 In the experiment, the test conditions employed AVS2 general, the RA configuration. 实验平台是Intel (R)Xeon(R)CPU X566002. 80GHZ 2. 79GHZ 23.96内存。 Experimental platform is Intel (R) Xeon (R) CPU X566002. 80GHZ 2. 79GHZ 23.96 memory. 表2表示采用本发明中的环路滤波方法,相比&1«*(^仏¥321©2.0^01)中的环路滤波方法的性能比较。 Table 2 shows the use of loop filtering method of the present invention, compared to 1 & «* (Fo ^ ¥ 321 © 2.0 ^ 01) compare the performance of the loop filtering method. 4kX2k、1080p、WVGA、WQVGA、720p分别表示分辨率为3840X2160、 1920Π 080、832Χ480、416Χ240、1280Χ720的测试序列。 4kX2k, 1080p, WVGA, WQVGA, 720p respectively resolution of 3840X2160, 1920Π 080,832Χ480,416Χ240,1280Χ720 test sequence. 编码性能的度量采用BD-rate,即在相同编码质量的情况下的码率节省。 Coding performance measure using BD-rate, i.e., saving in a case where the same bit rate encoding quality. 从实验结果来看,采用本发明实施例上述方案,在相同编码质量时,获得平均〇. 9%的码率节省。 From the experimental results, using the example of the embodiment of the present invention embodiment, when the same encoding quality, an average square. Bitrate savings of 9%.

[0036] 表2.性能比较(提出的滤波器设计vs RD2-ADI) [0036] Table 2. Performance comparison (proposed filter design vs RD2-ADI)

Figure CN103220529BD00071

[0038] 在编解码进行环路滤波时,通过利用编解码端块划分信息得到图像块(包括编码块、预测块和变换块)划分结构,环路滤波则在块(如编码块、预测块和变换块)边界进行, 若编码块的编码残差为〇(即CBP = 0),则仅在编码块和预测块边界进行滤波;滤波流程为预定义滤波分割块,把图像划分为滤波分割单元块(如16X16块大小、最大编码单元块大小或图像帧),可以预定义或在码流中说明滤波分割块的尺寸,为了统一滤波器设计,滤波顺序为依次对每个滤波分割块进行,先垂直边界后水平边界,即先对滤波分割块内的所有垂直边界进行滤波再对水平边界进行滤波;若边界包含编码块、预测块和变换块边界,则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策,若决策需要滤波该边界,则进行接下来的像素级滤波过程,否则跳过该边 [0038] When the codec performing loop filtering, to obtain an image block (including coding block, a transform block and a prediction block) divided structure, loop filtering at block (encoding block division information by using the codec end block, a prediction block and transform block) boundaries, when the encoded residual coded block is square (i.e., CBP = 0), only the coding block and a prediction filtering on the block boundary; filtering process to filter a predefined division block, dividing the image into a filter unit block (e.g., 16X16 block size, the maximum block size of the coding unit or image frames), or may be predefined dimension filtered sub-blocks in the bitstream, for the unified filter design, the filter order is sequentially divided each filtered block , the first vertical boundary horizontal boundary, i.e. first of all the vertical boundary in filtering the divided blocks are filtered and then the horizontal boundary filtering; if the boundary comprises a coding block, a prediction block and a transform block boundary, the utilization mode, motion information, quantization parameter and a residual other boundary decision stage filter, if the filter needs to decision boundary, for the next pixel level filtering process, otherwise, skip this edge 界滤波;对需要进行滤波的块边界处利用阈值和像素局部纹理特征进行像素级滤波决策以决定每个像素行是否需要滤波以及采用的滤波强度(BS);最后利用设计的滤波器执行像素级滤波,滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度。 Filtering sector; be in need of filtering at the block boundary pixel using the threshold and the local texture features filtered pixel level decision to determine whether each pixel row requires filtering and filter strength (BS) used; filter performs pixel-level design Finally filter, the filter can be a direction vertical to the boundary may be at other angles according to the local characteristics of the pixel or block mode and boundary.

[0039] 在编解码进行块边界环路滤波时,所述方法包括编解码在得到重构图像后进行环路滤波时,如图7给出了本发明专利支持的一种滤波流程其步骤如下: When [0039] the block boundary loop filtering in encoding and decoding, said method includes codec loop filtering performed after obtaining a reconstructed image, Fig. 7 presents a filtering process of the present invention is supported Patent comprises the following steps :

[0040] 初始化:利用编解码端的模式信息和块分割信息块得到当前重构图像帧的块划分方式,设置当前帧最小滤波单元,将图像划分为滤波分割块,设置起始滤波分割块序号N = 0〇 [0040] Initialization: using the codec mode information and the terminal block division information of the current block to obtain a reconstructed image block division manner of the frame, set the current minimum filter frame means, dividing the image into blocks filtering, filtering the divided set the starting block number N = 0〇

[0041] 按照扫描顺序依次对图像内每个滤波分割块N进行边界滤波; [0041] each filtered sequentially N of the divided image block boundary filtering in the scanning order;

[0042] 2. 1.先对滤波分割块的所有垂直边界,依次检测其是否为编码块、预测块或变换块边界,若CBP = 0,则检测是否为编码块、预测块边界;若是则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策。 [0042] 2. 1. First of all vertical block boundary filtering divided sequentially detecting whether the coding block, or prediction block transform block boundary, if the CBP = 0, it is detected whether the coding block, a prediction block boundary; if yes utilization mode, motion information, quantization parameter, etc., and a residual level filtering decision boundary.

[0043] 2. 2.对所有垂直边界,若边界级滤波决策需要滤波该垂直边界,则对该垂直边界执行像素级滤波强度决策以及像素级滤波。 [0043] 2. 2. If the decision boundary filtering stage filters the required vertical boundary is performed for all the vertical boundary pixel level of the vertical boundary filtering strength decision level and a pixel filtering.

[0044] 2. 3.再对滤波分割块的所有水平边界,依次检测其是否为编码块、预测块或变换块边界,若CBP = 0,则检测是否为编码块、预测块边界;若是则利用模式、运动信息、量化参数以及残差等进行边界级滤波决策。 [0044] 2. all horizontal boundary filtering again divided blocks sequentially detecting whether the coding block, or prediction block transform block boundary, if the CBP = 0, it is detected whether the coding block, a prediction block boundary; if yes utilization mode, motion information, quantization parameter, etc., and a residual level filtering decision boundary.

[0045] 2. 4.对所有水平边界,若边界级滤波决策需要滤波该水平边界,则对该水平边界执行像素级滤波强度决策以及像素级滤波。 [0045] 2. 4. When the decision boundary filtering stage filters the required horizontal boundary, the filter strength decision pixel level and pixel-level filtering is performed for all the horizontal boundary of the horizontal boundary.

[0046] 2. 5.判断是否整帧滤波完毕,若没有则滤波分割块序号N增加,跳到2. 1继续执行下一个滤波分割块的滤波。 [0046] 2. The filter determines whether the entire frame is completed, if not, the block division number N to increase the filtering, skip to the next 2.1 to continue filtering a divided block filtering.

[0047] 3.整帧滤波完毕; [0047] 3. The entire filter frame is completed;

[0048] 滤波分割块大小为16X16(这里仅是举例说明,具体大小可以预设值或在码流中说明),最小滤波单元为8X8块(如图6所示),需要利用边界处像素跳变和阈值表中的alpha beta以及像素局部跳变特征进行像素级滤波强度(BS)决策以及选择不同滤波强度下不同像素位置处的FIR滤波器(如附表2或附表3所示)。 [0048] block size is divided 16X16 filter (here are merely exemplary, or particular size may be preset values ​​described in the bit stream), the filter unit is a minimum 8X8 block (FIG. 6), the boundary pixels need to use hops and varying the threshold value table, and pixel alpha beta hopping local features pixel intensity level filtering (BS) and the decision of the FIR filter at different pixel positions with different filter strength (as shown in table 2 or table 3). 对于具体一个块边界的滤波过程如图8所示。 For a particular block boundary filtering process shown in Figure 8.

[0049] 滤波强度决策过程具体如下: [0049] The filter strength decision-making process as follows:

[0050] 如果abs (p0 - q0) <Alpha&&abs (p0 - q0) >1,则进行下面的步骤;否贝丨】,Bs = 0。 [0050] If abs (p0 - q0) <Alpha && abs (p0 - q0)> 1, the following steps are performed; NO shellfish] Shu, Bs = 0.

[0051] 设定左右平坦度变量FL和FR,并且两者的初值都设为0。 [0051] FL is set around the flatness and variable FR, and both are set to the initial value 0.

[0052] 如果abs (p0 - pi)〈Beta,则FL+ = 2 ; [0052] If abs (p0 - pi) <Beta, the FL + = 2;

[0053] 如果abs (p0 - p2)〈Beta,则FL++ ; [0053] If abs (p0 - p2) <Beta, the FL ++;

[0054] 如果abs (q0 - ql)〈Beta,则FR+ = 2 ; [0054] If abs (q0 - ql) <Beta, the FR + = 2;

[0055] 如果abs (q0 - q2)〈Beta,则FR++ ; [0055] If abs (q0 - q2) <Beta, the FR ++;

[0056] 左右平坦度之和FS = FL+FR。 [0056] around in flatness and FS = FL + FR.

[0057] 根据FS的值,做如下判断: [0057] The value of FS, is determined as follows:

[0058] 当FS = 6时,若p0 = pl,并且q0 = ql,则边界强度Bs = 4。 [0058] When FS = 6, if p0 = pl, and q0 = ql, then the boundary strength Bs = 4. 否则Bs = 3 ; Otherwise Bs = 3;

[0059] 当FS = 5时,若p0 = pl,并且q0 = ql,则边界强度Bs = 3。 [0059] When FS = 5, if p0 = pl, and q0 = ql, then the boundary strength Bs = 3. 否则Bs = 2 ; Otherwise Bs = 2;

[0060] 当FS = 4 时,若FL = 2,则Bs = 2。 [0060] When FS = 4, when FL = 2, then Bs = 2. 否则,Bs = 1。 Otherwise, Bs = 1.

[0061] 当FS = 3 时,并且abs (pi - ql)〈Beta,则Bs = 1。 [0061] When FS = 3, and abs (pi - ql) <Beta, then Bs = 1. 否则,Bs = 0。 Otherwise, Bs = 0.

[0062] 当FS为其它值时,Bs = 0。 [0062] When the other values ​​FS, Bs = 0.

[0063] 根据编码情况修正边界强度; [0063] The correction encoded boundary strength situation;

[0064] 若按照帧模式进行编码,并对色度分量进行滤波,Bs减1 : [0064] When the frame is encoded in accordance with the mode, filtering and chrominance components, Bs minus 1:

[0065] 若按照场模式进行编码,并满足下列条件之一,Bs减1 : [0065] When encoded as field mode, and the following conditions are satisfied, Bs is decremented by 1:

[0066] 对亮度分量在垂直方向上进行滤波; [0066] The filtered luminance component in the vertical direction;

[0067] 对色度分量进行滤波; [0067] filtering the chrominance component;

[0068] 当BS>0时,执行像素行的像素级滤波,滤波器采用附表2所示: [0068] When the BS> 0, the pixel row performing pixel-level filtering, using the filter shown in Table 2:

[0069] Bs等于4时的边界滤波过程; [0069] Bs equal to the boundary 4 of the filtering process;

[0070] 边界滤波强度Bs的值为4时,对p0、pl、p2、p3和q0、ql、q2、q3滤波的计算过程如下(P0、P1、P2和Q0、Q1、Q2是滤波后的值): [0070] When the boundary filtering strength Bs is 4, p0, pl, p2, p3, and q0, ql, q2, q3 filter calculated as follows (P0, P1, P2, and Q0, Q1, Q2 is filtered value):

[0071] PO = ((p0〈〈3)+p0+(p2〈〈2)+p2+(pl〈〈2) + (q0〈〈3)-q0+(ql〈〈2) + (q2〈〈l) +q2+16)»5 ; [0071] PO = ((p0 << 3) + p0 + (p2 << 2) + p2 + (pl << 2) + (q0 << 3) -q0 + (ql << 2) + (q2 << l) + q2 + 16) »5;

[0072] Pl = ((p0«2) + (p0«l) + (pl«l)+pl+(p2«2)+q0+(q0«l)+8)»4 ; [0072] Pl = ((p0 «2) + (p0« l) + (pl «l) + pl + (p2« 2) + q0 + (q0 «l) +8)» 4;

[0073] P2 = (p0+(pl«l)+p2+(p2«l)+p3+q0+4)»3 ; [0073] P2 = (p0 + (pl «l) + p2 + (p2« l) + p3 + q0 + 4) »3;

[0074] QO = ((q0〈〈3)+q0+(q2〈〈2)+q2+(ql〈〈2) + (p0〈〈3)-p0+(pl〈〈2) + (p2〈〈l) +p2+16)»5 ; [0074] QO = ((q0 << 3) + q0 + (q2 << 2) + q2 + (ql << 2) + (p0 << 3) -p0 + (pl << 2) + (p2 << l) + p2 + 16) »5;

[0075] Ql = ((q0«2) + (q0«l) + (ql«l)+ql+(q2«2)+p0+(p0«l)+8)»4 ; [0075] Ql = ((q0 «2) + (q0« l) + (ql «l) + ql + (q2« 2) + p0 + (p0 «l) +8)» 4;

[0076] Q2 = (q0+(ql«l)+q2+(q2«l)+q3+p0+4)»3 ; [0076] Q2 = (q0 + (ql «l) + q2 + (q2« l) + q3 + p0 + 4) »3;

[0077] Bs等于3时的边界滤波过程; [0077] Bs is equal to 3 when a boundary filtering process;

[0078] 边界滤波强度Bs的值为3时,对p0、pi和q0、ql滤波的计算过程如下(P0、Pl和QO、Ql是滤波后的值): [0078] boundary filtering strength Bs of the value of 3, for p0, pi calculation and q0, ql filtered follows (P0, Pl, and QO, Ql is the filtered value):

[0079] PO = (p2+(pl«2) + (pO«2) + (pO«l) + (qO«2)+ql+8)»4 ; [0079] PO = (p2 + (pl «2) + (pO« 2) + (pO «l) + (qO« 2) + ql + 8) »4;

[0080] QO = (pl+(pO«2) + (qO«2) + (qO«l) + (ql«2)+q2+8)»4 ; [0080] QO = (pl + (pO «2) + (qO« 2) + (qO «l) + (ql« 2) + q2 + 8) »4;

[0081] Pl = ((p2«l)+p2+(pK<3) + (p0«2)+q0+8)»4 ; [0081] Pl = ((p2 «l) + p2 + (pK <3) + (p0« 2) + q0 + 8) »4;

[0082] Ql = ((q2«l)+q2+(ql«3) + (qO«2)+pO+8)»4 ; [0082] Ql = ((q2 «l) + q2 + (ql« 3) + (qO «2) + pO + 8)» 4;

[0083] Bs等于2时的边界滤波过程; [0083] Bs is equal to 2 when a boundary filtering process;

[0084] 边界滤波强度Bs的值为2时,对p0和q0滤波的计算过程如下(PO和QO是滤波后的值): [0084] boundary filtering strength Bs is 2, the calculation of p0 and q0 filter follows (PO and QO are filtered value):

[0085] PO = ((pl«l)+pl+(p0«3) + (p0«l) + (q0«l)+q0+8)»4 ; [0085] PO = ((pl «l) + pl + (p0« 3) + (p0 «l) + (q0« l) + q0 + 8) »4;

[0086] QO = ((p0«l)+p0+(q0«3) + (q0«l) + (ql«l)+ql+8)»4 ; [0086] QO = ((p0 «l) + p0 + (q0« 3) + (q0 «l) + (ql« l) + ql + 8) »4;

[0087] Bs等于1时的边界滤波过程; [0087] Bs is equal to 1 the boundary filtering process;

[0088] 边界滤波强度Bs的值为1时,对p0和q0滤波的计算过程如下(PO和QO是滤波后的值): [0088] boundary filtering strength Bs value of 1, the calculation process of p0 and q0 filter follows (PO and QO are filtered value):

[0089] PO = ((p0«l)+p0+q0+2)»2 ; [0089] PO = ((p0 «l) + p0 + q0 + 2)» 2;

[0090] QO = ((q0〈〈l)+qO+pO+2)>>2。 [0090] QO = ((q0 << l) + qO + pO + 2) >> 2.

[0091] 本发明由于采用了上述技术方案,和已有的AVS标准中方法相比,具有如下创新和有益效果: [0091] The present invention adopts the above technical solutions, and conventional methods as compared to AVS standard, and has the following innovative advantages:

[0092] 第一、通过重用编解码端块划分信息得到图像块(包括编码块、预测块和变换块) 划分结构,环路滤波则在编码块、预测块和变换块边界进行,若编码块的CBP = 0,则仅在编码块和预测块边界进行滤波。 [0092] First, by reusing the codec end block to obtain an image block division information (including coding block, a transform block and a prediction block) divided structure, loop filtering is performed in the coding block, a prediction block and a transform block boundary, if the coding block the CBP = 0, then only the coding block and a prediction filtering on the block boundary. 这样可以提高了编码效率,而且减少了滤波的边界数目,从而降低了编码解码端复杂度,图4虚线显示AVSl原有的滤波方案在所有8X8变换块边界进行滤波,实线表示实际的编码块、预测块和变换块的划分方式,本方案只需要在实线处滤波, 从而滤波的边界数大大减少,例如对32X32的块,在所有8X8块边界滤波,则滤波边界数为32条8像素边界,而只在块边界滤波则只需要在8条8像素边界滤波,节省了4倍。 This can improve the coding efficiency, but also reduces the number of boundary filtering, thereby reducing the complexity of coding and decoding side, shown in dashed lines in FIG. 4 AVSl filtering scheme to filter the original 8X8 transform block boundaries in all, a solid line indicates an actual coding block dividing mode prediction block and a transform block, the present embodiment requires only a filtering solid line, so that the number of boundary filtering is greatly reduced, for example, block 32X32 in all 8X8 block boundary filtering, the filter number of boundary 32 8 pixels boundaries, but only in the block boundary filtering only need 8 8 pixel boundary filtering, saving 4 times. 需要滤波边界数和块尺寸大小关系如表1所示。 Filtering the block boundary and the number of required size relationship shown in Table 1. 所以在高分辨率视频中,当大块选择概率较多时, 采用本方法能显著减少需要滤波边界数,从而减少编解码端复杂度。 Therefore, high-resolution video, when the probability of more selected bulk, the present method can significantly reduce the required number of filtering boundaries, thereby reducing the complexity of encoding and decoding side.

[0093] 表1.需要滤波边界数目节省和块尺寸大小关系 [0093] Table 1. The number of required boundary filtering block size and save the relationship

Figure CN103220529BD00101

[0095] 第二、本方案预设值滤波分割块(如16X16块大小、最大编码单元块大小或图像帧,可以在码流中说明),在滤波之前,首先图像按照滤波分割块进行划分,然后按照扫描顺序依次对每个滤波分割块先垂直边界后水平边界的顺序滤波,这样能方便和统一滤波器设计,提高自适应性以及降低硬件电路的设计复杂度。 [0095] Second, the program defaults divided block filter (e.g., 16X16 block size, the maximum block size of the coding unit or image frames, may be described in the code stream), before the filter, the filter according to the first divided image division block, then sequentially scanning order sequentially divided block boundary filtering to the vertical and horizontal boundaries after each filtering, and so can easily unified filter design, improved adaptability and design complexity of the hardware circuit.

[0096] 第三、本发明方案的新的滤波阈值系数表如附表1和滤波器使用附表2所示,同时滤波方向可以根据模式信息和块边界局部图像特征来抉择,而不再固定为垂直块边界,可以更好提高编码效率和主观质量。 [0096] Third, the new filter coefficient of the threshold of the present invention, such as Table 1 and Table 2 Schedule filter uses, while the filtering direction can choose according to the local image feature mode information and the block boundary, instead of fixed vertical block boundary, can better improve the coding efficiency and subjective quality.

[0097] 本领域技术人员还可以了解到本发明实施例列出的各种说明性逻辑块(illustrative logical block),单元,和步骤可以通过电子硬件、电脑软件,或两者的结合进行实现。 [0097] Those skilled in the art that the present invention may also be listed in Example various illustrative logical blocks (illustrative logical block), means, and steps may be implemented by an electronic computer software, with hardware, or both. 为清楚展示硬件和软件的可替换性(interchangeability),上述的各种说明性部件(illustrative components),单元和步骤已经通用地描述了它们的功能。 To clearly show alternative hardware and software of (interchangeability), the above-described various illustrative components (illustrative components), common unit and steps have described their function. 这样的功能是通过硬件还是软件来实现取决于特定的应用和整个系统的设计要求。 This function is implemented by hardware or software depends upon the particular application and design requirements of the system. 本领域技术人员可以对于每种特定的应用,可以使用各种方法实现所述的功能,但这种实现不应被理解为超出本发明实施例保护的范围。 Those skilled in the art can for each particular application, various methods implement the described functionality, but such implementation should not be construed as beyond the scope of protection of the present invention.

[0098] 本发明实施例中所描述的各种说明性的逻辑块,或单元都可以通过通用处理器, 数字信号处理器,专用集成电路(ASIC),现场可编程门阵列(FPGA)或其它可编程逻辑装置,离散门或晶体管逻辑,离散硬件部件,或上述任何组合的设计来实现或操作所描述的功能。 [0098] The various illustrative logical blocks described in the embodiment of the present invention, or the unit may be a general purpose processor, a digital signal processor, application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to implement functions or operations described. 通用处理器可以为微处理器,可选地,该通用处理器也可以为任何传统的处理器、控制器、微控制器或状态机。 A general purpose processor may be a microprocessor, optionally, the general-purpose processor may be any conventional processor, controller, microcontroller, or state machine. 处理器也可以通过计算装置的组合来实现,例如数字信号处理器和微处理器,多个微处理器,一个或多个微处理器联合一个数字信号处理器核,或任何其它类似的配置来实现。 A processor may also be implemented by a combination of computing devices, such as a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors combined a digital signal processor core, or any other similar configuration achieve.

[0099] 本发明实施例中所描述的方法或算法的步骤可以直接嵌入硬件、处理器执行的软件模块、或者这两者的结合。 Steps of a method or algorithm described in the embodiments of the invention [0099] This may be directly embedded in the hardware, a software module executed by a processor, or a combination of both. 软件模块可以存储于RAM存储器、闪存、ROM存储器、EPROM 存储器、EEPROM存储器、寄存器、硬盘、可移动磁盘、⑶-ROM或本领域中其它任意形式的存储媒介中。 A software module may be stored in RAM memory, any other form of flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, ⑶-ROM, or a storage medium in the art. 示例性地,存储媒介可以与处理器连接,以使得处理器可以从存储媒介中读取信息,并可以向存储媒介存写信息。 Illustratively, the storage medium may be coupled to the processor such that the processor can read information from the storage medium, and may be stored writing information to the storage medium. 可选地,存储媒介还可以集成到处理器中。 Alternatively, the storage medium may be integral to the processor. 处理器和存储媒介可以设置于ASIC中,ASIC可以设置于用户终端中。 Processor and the storage medium may be provided in an ASIC, ASIC may be provided in a user terminal. 可选地,处理器和存储媒介也可以设置于用户终端中的不同的部件中。 Alternatively, the processor and the storage medium may be provided in a user terminal different components.

[0100] 在一个或多个示例性的设计中,本发明实施例所描述的上述功能可以在硬件、软件、固件或这三者的任意组合来实现。 [0100] In one or more exemplary designs, the functions of the above-described embodiment of the present invention described embodiments may be implemented in hardware, software, firmware or any combination of the three. 如果在软件中实现,这些功能可以存储与电脑可读的媒介上,或以一个或多个指令或代码形式传输于电脑可读的媒介上。 If implemented in software, the functions may be stored on computer-readable media, or transmitted as one or more instructions or code on a computer-readable form of the medium. 电脑可读媒介包括电脑存储媒介和便于使得让电脑程序从一个地方转移到其它地方的通信媒介。 Computer-readable media include computer storage media and so allow the computer to facilitate program from one place to the other parts of the communication medium. 存储媒介可以是任何通用或特殊电脑可以接入访问的可用媒体。 Storage medium may be any available media that can be accessed general purpose or special computer access. 例如,这样的电脑可读媒体可以包括但不限于RAM、ROM、EEPROM、CD-ROM或其它光盘存储、磁盘存储或其它磁性存储装置,或其它任何可以用于承载或存储以指令或数据结构和其它可被通用或特殊电脑、或通用或特殊处理器读取形式的程序代码的媒介。 For example, such computer-readable media can include but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other may be used to carry or store instructions or data structures and It may be other general purpose or special computer, or media program codes read a general purpose or special form of processor. 此外,任何连接都可以被适当地定义为电脑可读媒介,例如,如果软件是从一个网站站点、服务器或其它远程资源通过一个同轴电缆、光纤电脑、双绞线、数字用户线(DSL)或以例如红外、无线和微波等无线方式传输的也被包含在所定义的电脑可读媒介中。 Also, any connection may be suitably defined as a computer-readable medium, e.g., if software is transmitted from a website site, server, or other remote source through a coaxial cable, fiber optic computer, twisted pair, digital subscriber line (DSL) or to transmit infrared, radio, and microwave wireless manner, for example, it is also included in the computer readable medium as defined. 所述的碟片(disk)和磁盘(disc)包括压缩磁盘、镭射盘、光盘、DVD、 软盘和蓝光光盘,磁盘通常以磁性复制数据,而碟片通常以激光进行光学复制数据。 The disc (Disk) and disk (Disc) comprises a compact disc, laser disc, optical disc, DVD, floppy disk and blu-ray disc disks usually reproduce data magnetically, while discs usually reproduce data optically with a laser. 上述的组合也可以包含在电脑可读媒介中。 Combinations of the above may also be included in the computer readable medium.

[0101] 附表1本发明块边界阈值α和β与Index的关系 [0101] Schedule invention a block boundary threshold value α and β relationship with Index

Figure CN103220529BD00111

[0103] 附表2.本发明使用的滤波器参数设计 [0103] Table 2. The design parameters of a filter used in the present invention

Figure CN103220529BD00112

[0105] 附表3. AVS原有滤波器参数设计 [0105] Table 3. AVS original design parameters of the filter

Figure CN103220529BD00121

[0107] 本文中应用了具体实施例对本发明的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的一般技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处。 [0107] As used herein specific embodiments of the application of the principles and embodiments of the invention are set forth in the above described embodiments are only used to help understand the method and core idea of ​​the present invention; the same time, for those of ordinary skill in the art, according to the ideas of the present invention, there are modifications to the specific embodiments and application scope. 综上所述,本说明书内容不应理解为对本发明的限制。 Therefore, the specification shall not be construed as limiting the present invention.

Claims (7)

1. 一种视频编解码端的环路滤波的方法,其特征在于包括如下步骤: 步骤1)、利用编解码端块划分信息得到图像块划分结构,环路滤波则在所述图像块边界进行,若编码块的编码残差为0 (即CBP= 0),则仅在编码块和预测块边界进行滤波;滤波流程为预定义滤波分割块,把图像划分为滤波分割单元块,可以预定义或在码流中说明滤波分割块的尺寸,为了统一滤波器设计,滤波顺序为依次对每个滤波分割块进行,先垂直边界后水平边界,即先对滤波分割块内的所有垂直边界进行滤波再对水平边界进行滤波; 步骤2)、若边界包含编码块、预测块和变换块边界,则利用模式、运动信息、量化参数以及残差进行边界级滤波决策, 2. 1、若决策需要滤波该边界,则进行接下来的像素级滤波过程; 2. 2、否则跳过该边界滤波;对需要进行滤波的块边界处进行像素级 1. A method of loop filtering video codec side, comprising the following steps: Step 1), using the codec end block division information to obtain an image block division structure, loop filtering is performed on the block boundary image, If the encoded residual coding block is 0 (i.e., CBP = 0), only the filtered prediction block and the encoding block boundary; filtering process to filter a predefined division block, dividing the image into a filtering unit block, may be predefined or filtering the divided block size described in the bit stream, for the unified filter design, the filter order of sequentially dividing each filtered block, the horizontal boundary, i.e. first of all the vertical boundary filtering block segmentation and then filtered after the first vertical boundary horizontal boundary filtering; step 2), if the boundaries of the coding block comprises a prediction block and a transform block boundary, the utilization mode, motion information, quantization parameters and a residual level filtering decision boundary, 2.1, if the decision required filtering boundary, for the next pixel level filtering process; 2.2, otherwise skip the boundary filtering; at the pixel level block boundary filtering is required for 波决策以决定每个像素行是否需要滤波以及采用的滤波强度(BS); 步骤3)、最后执行像素级滤波,滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度,其中参与图像滤波的块结构,包括但不限于编码块、预测块和变换块;滤波分割块为方便滤波操作,而对图像进行统一尺寸的划分方式,对图像可以按照16X16尺寸滤波分割块划分。 Wave decision to determine whether each pixel row requires filtering and filter strength (BS) used; step 3), the last stage performing pixel filter, the filter may be a direction vertical to the boundary may be at other angles according to the local characteristics of the pixel or block mode and boundary wherein involved in the block structure of the image filtering, including but not limited to, the coding block, a prediction block and a transform block; filtering divided blocks to facilitate filtering operation, and the image division manner a uniform size, the image can follow 16X16 size filtering the divided blocks divided .
2. 根据权利要求1所述的一种视频编解码端的环路滤波的方法,其特征在于,所述步骤1)还包括:图像块包括编码块、预测块和变换块。 The loop filtering method according to a video codec terminal 1 according to claim wherein said step a) further comprises: an image block comprises a coding block, a prediction block and a transform block.
3. 根据权利要求1所述的一种视频编解码端的环路滤波的方法,其特征在于,所述步骤1)还包括:环路滤波发生的边界在所述图像块边界,当编码块的编码残差为零时,即CBP =〇,则只在编码块和预测块边界进行滤波,同时定义滤波的最小尺寸8X8块大小;编码块的尺寸包含了从64X64到8X8大小,根据RDO选择,采用四叉树迭代递归划分,是编码处理的基本单元;预测块为在编码块的基础上的进一步根据预测特性的进一步划分;变换块也是在编码块的基础上根据残差的分布特性的进一步划分。 A method according to a codec end loop filtering video according to claim 1, wherein said step a) further comprises: a boundary loop filtering occurs in the image block boundary, when the coding block when the residual coding is zero, i.e., CBP = square, only in the encoding block and the prediction filter block boundaries, while the definition of the minimum size of the filtering 8X8 block size; size of the coding block contains from 8X8 to 64X64 size, according to the selection RDO, quadtree partitioning recursive iteration, is the basic unit of encoding processing; prediction block is further subdivided on the basis of a coding block according to the predicted characteristic; transform block is further based on the encoded block according to the distribution characteristic of the residual division.
4. 根据权利要求1所述的一种视频编解码端的环路滤波的方法,其特征在于,所述步骤1)还包括:对于整帧图像划分为滤波分割单元块,可以预定义或者在码流中说明滤波分割块的尺寸,按照扫描依次对每个滤波分割块进行滤波,先对垂直边界,再水平边界,检测边界是否包含编码块、预测块和变换块边界,若包含,则滤波该边界,否则跳过滤波该边界。 4. A video loop filtering method according to a codec as claimed in claim end, wherein said step a) further comprises: for filtering the whole frame is divided into block dividing unit, or may be predefined in the code dimension filtered stream sub-blocks, each filtered sequentially filtering the divided block according to a scan, the first vertical boundary, a horizontal boundary and then, whether the boundary comprises detecting a coding block, a prediction block and a transform block boundary, if included, the filter border, otherwise skip filtering the boundary.
5. 根据权利要求1所述的一种视频编解码端的环路滤波的方法,其特征在于,所述步骤2)还包括:根据滤波边界两边块的模式、运动信息、量化参数以及残差来决定当前块边界是否需要滤波,若边界级决策需要滤波该边界,则进行接下来的像素级滤波过程,否则跳过该边界滤波,若块边界两边块P和Q有帧内预测模式,则需要执行滤波;若P、Q块均为P 帧的帧间预测块,其残差均为〇,参考帧相同且运动矢量各分量小于一个像素,则决策该块边界不需要滤波,否则均需要滤波。 The loop filtering a video codec end method according to claim 1, wherein said step 2) further comprising: a filter according to the pattern on both sides of the boundary block, motion information, quantization parameters, and residuals determines whether the current block boundary filtering, if required filtering the boundary level decision boundary, for the next pixel level filtering process, or the boundary filtering is skipped, if both sides of the block boundary blocks P and Q have the intra prediction mode, it is necessary performs filtering; if P, Q are inter-predicted block P frames, which are square residual, the identical reference frame and the motion vector of each component is smaller than a pixel, then the filtering decision block boundary does not need, or are required filtering .
6. 根据权利要求1所述的一种视频编解码端的环路滤波的方法,其特征在于,所述步骤2)还包括:滤波决策需要滤波的块边界每行像素,利用边界两边像素的跳变和两个块平均QP值得到的索引所对应的阈值表对应值比较,来选择滤波强度;像素级滤波强度取值范围0-4,BS= 0表示不需要滤波,BS= 4,表示较为平坦区域,需要进行较强的平滑滤波。 6. A video loop filtering method according to a codec as claimed in claim end, wherein said step 2) further comprises: filtering decision block boundary filtering pixels per line required by both sides of the boundary pixel jumping comparison of the two blocks and the average QP becomes worth corresponding to the index table corresponding to the threshold value, to select the filter strength; pixel intensity level filtering in the range 0-4, BS = 0 indicates that no filtering, BS = 4, represents more flat areas, the need for a strong smoothing filter.
7. 根据权利要求1所述的一种视频编解码端的环路滤波的方法,其特征在于,所述步骤3)还包括:滤波方向可以垂直于边界也可以根据局部像素特性或块模式和边界成其他角度;选择不同滤波强度下不同像素位置处的FIR滤波器进行滤波;同时可以利用局部图像的纹理特性和边界两边块的模式信息决定滤波的方式是垂直于边界或者和边界成一定角度。 7. The method of claim 1 loop filtering in a video codec as claimed in claim end, wherein said step 3) further comprising: a filtering direction may be perpendicular to the boundary pixel characteristic may be partially or block mode in accordance with the boundary and other angles; the FIR filter at different pixel positions in different filtering filter strength; texture features while utilizing partial images on both sides of the block boundary and the mode information determining the filtering approach is vertical to the boundary or border and angled.
CN201310128415.2A 2013-04-15 2013-04-15 A video codec loop filtering implemented method CN103220529B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310128415.2A CN103220529B (en) 2013-04-15 2013-04-15 A video codec loop filtering implemented method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310128415.2A CN103220529B (en) 2013-04-15 2013-04-15 A video codec loop filtering implemented method

Publications (2)

Publication Number Publication Date
CN103220529A CN103220529A (en) 2013-07-24
CN103220529B true CN103220529B (en) 2016-02-24

Family

ID=48817933

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310128415.2A CN103220529B (en) 2013-04-15 2013-04-15 A video codec loop filtering implemented method

Country Status (1)

Country Link
CN (1) CN103220529B (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104754363B (en) * 2013-12-31 2017-08-08 展讯通信(上海)有限公司 Hevc loop filtering for the method and apparatus, an encoder and a decoder
CN103888764B (en) * 2014-03-14 2017-02-15 西安交通大学 A system and method for adaptive distortion-compensated video compression
CN104113765A (en) * 2014-07-28 2014-10-22 北京大学深圳研究生院 Video coding and decoding method and device
WO2016015198A1 (en) * 2014-07-28 2016-02-04 北京大学深圳研究生院 Video coding/decoding method and apparatus
CN105791865A (en) * 2014-12-22 2016-07-20 江苏省电力公司南京供电公司 Intra-frame prediction and deblocking filtering method
CN104683820B (en) * 2015-02-28 2018-03-06 华为技术有限公司 One kind of a loop filter and a loop filtering method
CN109845266A (en) * 2016-10-14 2019-06-04 联发科技股份有限公司 To remove the smoothing filtering method and device of ripple effect
CN107172423B (en) * 2017-03-31 2018-08-21 武汉斗鱼网络科技有限公司 Method and apparatus for filtering a video frame
CN107801024A (en) * 2017-11-09 2018-03-13 北京大学深圳研究生院 Boundary filtering method for intra-frame prediction

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1652610A (en) * 2005-01-31 2005-08-10 浙江大学 Deblocking filtering method and apparatus of video frequency or image
CN101841722A (en) * 2010-06-08 2010-09-22 上海交通大学 Detection device of filtering boundary strength and detection method thereof
CN102387366A (en) * 2005-03-18 2012-03-21 夏普株式会社 Methods and systems for extended spatial scalability with picture-level adaptation
CN102577377A (en) * 2009-08-04 2012-07-11 三星电子株式会社 Apparatus and method for deblocking filtering image data and video decoding apparatus and method using the same
CN102792692A (en) * 2010-01-14 2012-11-21 三星电子株式会社 Method and apparatus for encoding video by using deblocking filtering, and method and apparatus for decoding video by using deblocking filtering
CN102934429A (en) * 2010-05-18 2013-02-13 索尼公司 Image processor and image processing method
CN102948146A (en) * 2010-04-13 2013-02-27 三星电子株式会社 Video encoding method and video encoding apparatus and video decoding method and video decoding apparatus, which perform deblocking filtering based on tree-structure encoding units
CN103004197A (en) * 2010-04-23 2013-03-27 吴秀美 Image encoding device and method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1652610A (en) * 2005-01-31 2005-08-10 浙江大学 Deblocking filtering method and apparatus of video frequency or image
CN102387366A (en) * 2005-03-18 2012-03-21 夏普株式会社 Methods and systems for extended spatial scalability with picture-level adaptation
CN102577377A (en) * 2009-08-04 2012-07-11 三星电子株式会社 Apparatus and method for deblocking filtering image data and video decoding apparatus and method using the same
CN102792692A (en) * 2010-01-14 2012-11-21 三星电子株式会社 Method and apparatus for encoding video by using deblocking filtering, and method and apparatus for decoding video by using deblocking filtering
CN102948146A (en) * 2010-04-13 2013-02-27 三星电子株式会社 Video encoding method and video encoding apparatus and video decoding method and video decoding apparatus, which perform deblocking filtering based on tree-structure encoding units
CN103004197A (en) * 2010-04-23 2013-03-27 吴秀美 Image encoding device and method
CN102934429A (en) * 2010-05-18 2013-02-13 索尼公司 Image processor and image processing method
CN101841722A (en) * 2010-06-08 2010-09-22 上海交通大学 Detection device of filtering boundary strength and detection method thereof

Also Published As

Publication number Publication date
CN103220529A (en) 2013-07-24

Similar Documents

Publication Publication Date Title
Fu et al. Sample adaptive offset in the HEVC standard
JP5344636B2 (en) Simplified deblock filtering for reducing memory accesses and computational complexity
JP5396478B2 (en) Video coding method using adaptive loop filter
CN100566411C (en) Method, medium, and filter removing a blocking effect
KR101618021B1 (en) Video coding using a subset of intra prediction modes and corresponding directional transforms
CN103238323B (en) Decoding the position of the last significant coefficient within the block based on the scanning order of video blocks in the video coding
CN102204251B (en) Video coding using 4x4 and 8x8 transform is greater than the
CN104902283B (en) Video decoding method
CN104221373B (en) Coding for SAO / or signaling apparatus and method, and
KR101661828B1 (en) Adaptive overlapped block motion compensation
CN103999460B (en) Method and apparatus in a transform coding significant coefficients skip mode information
CN101711481B (en) Method and apparatus for video coding using prediction data refinement
CN104270639B (en) Provide a method and apparatus for reduced resolution update mode multiview video coding
KR101323546B1 (en) Filter prediction based on activity metrics in video coding
ES2681209T3 (en) Acceleration techniques for optimized quantization distortion rate
CN100345450C (en) Deblocking filtering method and apparatus of video frequency or image
CN103370936A (en) Method and apparatus for improved in-loop filtering
KR101864910B1 (en) Pixel level adaptive intra-smoothing
CN101371585B (en) Method and apparatus for providing reduced resolution update mode for multi-view video coding
CN102057676B (en) Method and system for video coder and decoder joint optimization
CN103636215A (en) Applying non-square transforms to video data
CN1929611A (en) Methods and apparatus for video intraprediction encoding and decoding
CN103392339B (en) A multi-measurement filtering method and apparatus
CN103688541A (en) Buffering prediction data in video coding
CN103348651A (en) Method and apparatus of transform unit partition with reduced complexity

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model