Research On HEVC Intra Coding Optimization Algorithm

With the development of video capture and display technologies, high and ultra high definition video applications are emerging. The video applications are developing to diversification and high-definition. All of these put forward higher requirements for video compression. In order to meet the requirements of video compression, the ITU-T Video Coding Experts Group(VCEG) and the Moving Picture Experts Group(MPEG) jointly formed the Joint Collaborative Team on Video Coding(JCT-VC). Aiming to establish and develop the next generation video coding standard, namely High Efficiency Video Coding (HEVC).Compared to H.264/AVC, HEVC can reduce bitrate requirements by 50% with comparable video quality. Although HEVC provides greater coding compression efficiency, it also leads to higher encoding complexity. Therefore, HEVC intra-coding optimization will become a research topic. We focus on the HEVC intra-coding optimization in this thesis. The main contributions of this thesis are as follows:(1) This chapter introduces two JND models in detail:JND model based on pixel domain and JND model based on DCT domain. We introduce JND model based on pixel domain to the next generation video coding standard (HEVC), and apply to the fast Coding Unit (CU) partition in HEVC intra coding. The algorithm effectively reduces the complexity of the Coding Unit (CU) partition.(2) The Coding Unit (CU) partition is based on the rate distortion optimization bottom-up pruning algorithm. The algorithm can make the Coding Unit (CU) structure to reach local optimum. But the exhaustive recursive manner has enormous encoding complexity. In this chapter, we proposed an image content based on fast Coding Unit (CU) partition algorithm which uses the video image feature and combines with the coding bits of each CU. The algorithm effectively reduces the complexity of the Coding Unit (CU) partition.(3) On the basis of H.264/AVC, HEVC adopted a more flexible block structure and diversity of prediction modes. HEVC supported size of Predict Unit (PU) from 4x4 to 64x64. For a variety of Prediction Unit (PU) of luminance, HEVC provides up to 35 prediction modes, which includes DC mode, Planar mode and 33 directional modes. This flexible block structure and prediction modes can match the complex texture feature of video image, reduce the forecast error and remove the spatial redundancy. But it also leads tremendous encoding complexity. In the chapter, we propose a fast intra mode decision method, which uses correlations of modes and the statistical results of the intra prediction modes. The algorithm effectively reduces the complexity of the intra mode decision.