CN107105297B - A fast optimization method for intra-frame prediction coding of 3D-HEVC depth map - Google Patents

Abstract

The invention discloses a kind of fast Optimizations for 3D-HEVC depth map intraframe predictive coding, it include: that the sum of its pixel variance and diagonal pixel absolute value of the difference are calculated to present encoding block CU, according to the sum of the absolute value of the variance of present encoding block and diagonal pixel value difference given threshold, by threshold value comparison, the depth for determining whether to terminate current CU in advance is divided；According to the sum of current prediction block PU outermost ring of pixels absolute value of the difference, whether given threshold belongs to smooth type by threshold value comparison current prediction block PU, to skip SDC coding, further decreases computation complexity.The present invention can be effectively reduced depth map intraframe predictive coding computation complexity under the premise of keeping 3D-HEVC code efficiency.

Description

A fast optimization method for intra-frame prediction coding of 3D-HEVC depth map

technical field

The invention relates to the field of video coding, in particular to a fast optimization method for intra-frame predictive coding of a 3D-HEVC depth map.

Background technique

With the rapid development of multimedia communication technology and various video terminal processing capabilities, 3D video is becoming more and more popular in life. Considering that the previous generation of multi-view video coding standards based on H.264 cannot meet the high-efficiency compression of the current increasing 3D video data volume, the stereoscopic video coding joint group JCT-3V has formulated a new generation of multi-view video coding standards 3D-HEVC (3D- High Efficiency Video Coding).

Although 3D-HEVC achieves high coding efficiency, it has the problem of high computational complexity, which seriously affects its practical application. Currently, the common 3D video format is texture video plus depth video format (Video Plus Depth), in which the depth map plays an important role in virtual viewpoint synthesis. However, the depth map coding process also introduces a large computational complexity, and its main time-consuming modules include intra-frame prediction coding and inter-frame prediction coding.

For intra-frame prediction coding, the existing 3D-HEVC uses a recursive coding block CU partition algorithm and numerous PU prediction units. At the same time, when calculating the rate-distortion cost value RD-cost for the depth map intra prediction unit, it is necessary to perform various encoding operations on the prediction unit, including SDC encoding and non-SDC encoding. If SDC coding is performed, the prediction residual of the current prediction block does not need to be transformed and quantized, and one or two values are directly used to replace the prediction residual value of the current prediction block according to the Depth Lookup Table (DLT). The SDC encoding also introduces a certain computational complexity while improving the encoding performance. Therefore, how to reduce the computational complexity of depth map intra-frame prediction coding while maintaining 3D-HEVC coding efficiency is an urgent problem to be solved.

Contents of the invention

The main purpose of the present invention is to propose a fast optimization method for intra-frame predictive coding of 3D-HEVC depth map in view of the large computational load caused by the recursive process of block segmentation recursive process and SDC coding process of 3D-HEVC depth map , effectively reducing the computational complexity of the encoder while ensuring that the encoding efficiency is almost unchanged.

The present invention adopts following technical scheme:

A fast optimization method for intra-frame predictive encoding of 3D-HEVC depth maps, characterized in that it comprises the following steps:

1) Calculate the pixel variance _D and the sum SD of the absolute value of the diagonal pixel difference for the current coding block CU;

2) Compare the pixel variance _{D and the sum SD of the absolute value of the diagonal pixel difference with the preset thresholds T D and} T _SD respectively to determine whether to skip the CU segmentation; if not, go to step 3); if so, then Go to step 4);

3) Judging whether the current coding block CU reaches the maximum CU depth, if so, then perform CU segmentation, and return to step 1); if not, enter step 4);

4) Calculate the sum S _P of the absolute value of the pixel difference in the outer circle of the current prediction block PU and compare it with the preset threshold T _P to determine the predictive coding mode.

Preferably, in step 1), the pixel variance D is calculated as follows:

Wherein Xi represents the _i -th pixel value of the current coding unit CU, and M represents the total number of pixels in the current coding unit CU.

Preferably, in step 1), the sum _SD of the absolute value of the diagonal pixel difference is calculated as follows:

Wherein, X _i,i represents the pixel value of row i, column i of the current coding unit CU, Xi _{, Ni} represents the pixel of row i, column Ni of the current coding unit CU, and N represents the size width value of the current coding unit CU.

Preferably, the pixel variance _{D and the absolute value SD of the diagonal pixel difference described in step 2) are compared and judged with the preset threshold T D and} T _SD respectively, and its judgment condition is:

D＞T _D ||S _D ＞T _SD .

Preferably, in step 4), the calculation formula of the sum S _P of the absolute values of the outer circle pixel differences of the current prediction block PU is:

Wherein, X _1,j represents the i-th pixel value in the first row of the current prediction unit PU, and X _K,j represents the i-th pixel in the K-th row of the PU. X _i,1 represents the pixel in row i, column 1 of the PU, X _i,K represents the pixel in row i, column K of the PU, and K represents the size of the current prediction unit PU.

Preferably, in step 4), the sum S _P of the absolute value of the difference between the pixels in the outer circle of the current coding block CU is calculated and compared with the preset threshold T _P for judgment. If S _P > T _P , SDC coding and non- SDC encoding, otherwise adopt SDC encoding.

Preferably, in step 4), the T _P is obtained by performing mean value calculation after sampling statistics of the current prediction block PU when performing SDC encoding.

As can be seen from the above description of the present invention, compared with the prior art, the present invention has the following beneficial effects:

1. The method of the present invention uses the variance of the coding block of the depth map and the image characteristics of the diagonal pixel difference to perform threshold judgment, skips some CUs that do not need to be divided, and saves coding time simply and effectively.

2. The method of the present invention makes full use of the smoothness of the coding block of the depth map to skip unnecessary non-SDC coding, and effectively reduces the computational complexity.

Description of drawings

Fig. 1 is a flowchart of the present invention.

Detailed ways

The present invention will be further described below through specific embodiments.

Referring to Fig. 1, a fast optimization method for 3D-HEVC depth map intra-frame prediction coding according to the present invention optimizes and improves coding unit segmentation and prediction unit SDC coding. The specific implementation steps are as follows:

1) For the coding block CU of the current input depth map, calculate its pixel variance _D and the sum SD of the absolute values of diagonal pixel differences.

The formula for calculating the pixel variance D is as follows:

Wherein Xi represents the _i -th pixel value of the current coding unit CU, and M represents the total number of pixels in the current coding unit CU.

The calculation formula of the absolute value _SD of the diagonal pixel difference of the current coding block CU is as follows:

Wherein, X _i,i represents the pixel value of row i, column i of the current coding unit CU, Xi _{, Ni} represents the pixel of row i, column Ni of the current coding unit CU, and N represents the size width value of the current coding unit CU.

2) Set double thresholds T _D and T _SD , compare the threshold with variance _D and the sum SD of absolute values of diagonal pixel differences, the formula for threshold judgment is:

D＞T _D ||S _D ＞T _SD ;

Among them, T _D and T _SD are the set thresholds, and the selection of the thresholds is obtained by experiment statistics.

If the above threshold judgment formula is not satisfied, go to step 4); if the above threshold judgment formula is satisfied, go to step 3).

3) Judging whether the current coding block CU reaches the maximum CU depth, if so, perform CU segmentation; if not, proceed to step 4).

4) Calculate the sum SP of the absolute value of the outer circle pixel difference of the current prediction block _PU , the calculation formula is as follows:

Wherein, X _1,j represents the i-th pixel value in the first row of the current prediction unit PU, and X _K,j represents the i-th pixel in the K-th row of the PU. X _i,1 represents the pixel in row i, column 1 of the PU, X _i,K represents the pixel in row i, column K of the PU, and K represents the size and width of the current prediction unit PU. A prediction block PU is an HEVC basic prediction unit. In intra-frame predictive coding, the current 2N×2N coding block CU may have two predictive block PUs of 2N×2N and N×N sizes.

5) Set the threshold T _P , compare _SP and T _P , and determine whether to skip SDC coding in the end. The threshold judgment formula is:

S _P > T _P ;

Where T _P is a set threshold, and the setting of the threshold is obtained by performing mean value calculation after sampling and statistics of the current prediction block PU when performing SDC encoding. If the judgment formula is satisfied, SDC coding and non-SDC coding are performed on the prediction block PU; otherwise, it belongs to the smooth type, and SDC coding is performed on the prediction block.

The above is only a specific embodiment of the present invention, but the design concept of the present invention is not limited thereto, and any insubstantial changes made to the present invention by using this concept should be an act of violating the protection scope of the present invention.

Claims (2)

1. A fast optimization method for 3D-HEVC depth map intra-frame predictive coding, it is characterized in that, comprising the steps: 1) Calculate the pixel variance _D and the sum SD of the absolute value of the diagonal pixel difference for the current coding block CU; the sum _SD of the absolute value of the diagonal pixel difference is calculated as follows: Wherein, X _i,i represents the pixel value of the i-th row and i-th column of the current coding unit CU, Xi _{, Ni} represents the pixel of the i-th row and Ni-th column of the current coding unit CU, and N represents the size width value of the current coding unit CU; 2) Compare the pixel variance _{D and the absolute value SD of the diagonal pixel difference with the preset thresholds T D and} T _SD respectively to determine whether to skip the CU segmentation, and the judgment conditions are: D＞T _D ||S _D ＞T _SD ; if the judgment condition is met, go to step 3); if the judgment condition is not met, go to step 4); 3) Judging whether the current coding block CU reaches the maximum CU depth, if so, then perform CU segmentation, and return to step 1); if not, enter step 4); 4) Calculate the sum S _P of the absolute value of the pixel difference in the outer circle of the current prediction block PU and compare it with the preset threshold T _P to determine the predictive coding method. The calculation formula of S _P is: Among them, X _1,j represents the value of the i-th pixel in the first row of the current prediction unit PU, X _K,j represents the i-th pixel in the K-th row of the PU, and X _i,1 represents the pixel in the i-th row and the first column of the PU , Xi _{, K} represents the i-th row and K-column pixel of the PU, and K represents the size of the current prediction unit PU; If S _P > T _P , then use SDC coding and non-SDC coding, otherwise use SDC coding, where T _P is obtained by performing mean value calculation after SDC coding of the current prediction block PU. 2. a kind of fast optimization method for 3D-HEVC depth map intra-frame predictive coding as claimed in claim 1, is characterized in that, in step 1), described pixel variance D is calculated as follows: Wherein Xi represents the _i -th pixel value of the current coding unit CU, and M represents the total number of pixels in the current coding unit CU.