Research On Intra Coding Algorithm For H.265/HEVC

With the popularity of high definition and ultra high definition video application, the video coding standard also need for innovation, and a new video coding standard, high efficient video coding (HEVC) arises at the historic moment. HEVC employs a lot of new technologies, and the new technologies have greatly improved the coding efficiency. At the same time, the technologies have also increased the coding complexity. In some cases where we need to consider the real-time requirements, the complexity will greatly limit the application of HEVC. Thus, how to reduce the HEVC intra coding complexity on the premise of guaranteeing coding performance has become important.This paper firstly introduces the history of video coding standard and the key technical points of HEVC. Secondly, the intra coding algorithm of HEVC is studyed. The paper analyzes the intra coding algorithm and its complexity, and summarizes the domestic and foreign research status about intra coding fast algorithm. Finally, two fast intra coding algorithms are proposed. The major contributions of this paper are as follows:(1) We propose a fast block partitioning algorithm which is based on spatial correlation and image texture. Firstly, a concept of most possible depth range (MPDR) is proposed according to the correlation of adjacent CTUs. Secondly, the difference between the current CTU texture and neighboring CTU texture is analyzed by detecting the dominant edge direction. According to the texture difference, the depth range of the current CTU is chose adaptively. Finally, skip some unit coding calculation according to the texture complexity which is detected by the pixel variance. Compared with HM 13.0, experimental results reveal that the proposed algorithm can save about 20% computational complexity on average whereas the average bit rate increment is 0.84% and the decrement of PSNR is 0.04 dB.(2) We propose a fast mode decision algorithm for HEVC intra prediction which is based on dominant edge assent (DEA) and its distribution. Firsly, classify all the 35 intra prediction modes into 4 type in accordance with the dominant edge. The mode type of the current coding block is decided by the minimum DEA. Secondly, determine whether the texture of the current block is obvious directional or smooth according to the distribution of DEA, and then choose the different number of modes as candidate modes for the RMD process. Finally, the size-based number of RDO candidate modes is modified. Compared with HM 13.0, the experimental results reveal that the proposed algorithm can save 36.26% coding time on average with a negligible loss of coding efficiency.(3) Further more, we combine the above proposed fast block partition and mode decision algorithms. Employ the proposed fast block partitioning algorithm to do the block partition and choose the best prediction mode for a PU by using the proposed fast mode decision algorithm. The combined algorithm can save 46.11% coding time on average with small coding efficiency loss.