Research On 3D-HEVC Depth Intra Coding Fast Algorithm

With the advent of the multimedia age,pictures and videos began to be widely used by people.With the development of science and technology,people's demands for video quality have been continuously improved.Previous video coding standards have gradually failed to meet the fast and efficient processing of batch high-definition video.High-definition and ultra-clear video contain a huge amount of data,full transmission makes it difficult to bear on the bandwidth of the network.Video coding is to compress and encode video,eliminating the time and space redundancy in the data as much as possible,which makes the video transmission more simple and efficient.3D-High Efficiency Video Coding(3D-HEVC)is derived from HEVC.It is can not only complete the previous high-definition video processing,but also efficiently handle 3D high-definition video,which is the latest video coding standard.Although the related coding standard achieves high coding efficiency,it also increases the computational complexity accordingly,which limits its use in practical applications.This paper bases on the inheritance of the coding parent and child layers and the mode decision to make depth intra prediction fast algorithm with low complexity.In the 3D-HEVC depth intra encoding process,since the encoding standard uses a quad-tree-based hybrid encoding,the depth intra encoding mode decision will be repeated both at the root node and at the child nodes.Although this operation guarantees the encoding performance,it also results in an increase in computational complexity.According to the inheritance of parent and child layers,this paper presents a depth intra fast algorithm.The parent layer is prioritized by the sub-layer coding,and the number of sub-layer candidate modes is limited according to the best mode obtained by the completed parent layer,skipping the insignificant coding mode,thereby reducing the computational complexity.In addition,in the best mode decision,according to the first intra mode of the candidate list,decide whether to terminate the DMM mode decision in advance.Experimental results show that the proposed fast algorithm can save coding time effectively.This paper also implements a mode decision algorithm based on the idea of threshold.Through the experimentally derived correlation threshold,the coding block is divided into smooth blocks and motion blocks.The smooth block refers to the fact that the object in the coding block is basically mot ionless,and the conventional coding technology only needs to use HEVC in the coding process,and the motion block relatively has a richer texture,and the new technology in the 3D-HEVC coding standard needs to be used.Through several comparisons,different blocks use different coding strategies.Skipping the redundant modes ahead of time does not require traversing all the modes and reducing the number of traversal times.Experimental results show that the depth intra fast algorithm can effectively reduce the computational complexity of the encoder.So far,the fast algorithm for depth map coding has been flourishing.The research work in this paper is only concerned with the small aspects of intra prediction coding.There are many techniques in the coding mode that await the researcher to optimize,hoping to promote the further development of depth intra coding.