Research On High Efficiency Video Coding Method Based On 3D-HEVC

In recent years,with the continuous development of multimedia technology,3D video has gradually attracted people's attention.3D video brings great technical requirements to the audience,but also brings higher technical requirements,such as large amount of encoded data,high storage requirements,and difficult transmission.In order to better meet the needs of the masses and realize the popularization of 3D video,the Stereo Video Coding Joint Working Group proposed a new generation of high-efficiency stereo video coding standard,namely 3D-HEVC,based on HEVC.3D-HEVC adopts the format of multi-view and depth map,and utilizes the correlation between viewpoints to reduce redundant information between viewpoints.Although 3D-HEVC has greatly reduced the redundant information of 3D video in the encoding process,the complexity of the encoding algorithm is still high,which seriously affects the popularity of 3D-HEVC coding standards.After analyzing various technologies introduced by 3D-HEVC and 3D-HEVC coding structure,this paper optimizes and improves the following four-tree recursive partitioning process and prediction unit intra-mode selection process:(1)The technique of recursive partitioning of coding unit quadtree in 3D-HEVC and the basic principles and related algorithms of intra-frame coding are elaborated.The advantages and disadvantages of these methods are analyzed and compared,and the association between depth map and texture map is analyzed.Sex.(2)Aiming at the high complexity of the 3D-HEVC depth map in the recursive partitioning process of quadtree,the machine learning technique is introduced,and the Classification and Regression Tree(CART)is applied to the quadtree recursive partitioning process.The model predicts the optimal partition depth of the current Largest Coding Unit(LCU)and skips the DMM pattern search process according to the prediction result,thereby terminating the exhaustive quadtree recursive partitioning process and the DMM search process in advance.The experimental results show that the CART-based adaptive depth decision algorithm proposed in this paper can reduce the depth map coding time by 25.5% on average while reducing the BDBR by 0.51%,which can effectively reduce the video coding quality.The coding complexity of the depth map quadtree partitioning process.(3)Based on the analysis of the characteristics of coding blocks processed by each intra mode in 3D-HEVC,a fast algorithm for intra-mode selection based on gradient information is proposed.The algorithm uses the gradient information of the coding block to divide the current coding block into weak gradient block,strong gradient block and non-weak non-strong gradient block,and selects the appropriate intra mode to be added to the candidate list according to the characteristics of different gradient blocks.The experimental results show that the proposed algorithm can reduce the coding time by 30.6% on average,while the BDBR only increases by 0.66%,which effectively reduces the algorithm complexity of the encoder intra prediction process.