Font Size: a A A

Research On Improving The Video Coding Performance Of HEVC

Posted on:2013-07-15Degree:MasterType:Thesis
Country:ChinaCandidate:L J KangFull Text:PDF
GTID:2248330395956615Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Collaborative Team on Video Coding (JCT-VC) is currently developing a new generation video coding standard, HEVC (High Efficiency Video Coding), to enhance the coding efficiency. The goal of the standard is to achieve one time higher compression capability than the H.264/AVC High Profile. Now, as HEVC standardization process is being promoted, its video coding efficiency can be dramatically increased, which means approximately40%bit-rate decrease compared with H.264/AVC at the same video quality. During the development of HEVC, this thesis follows up and explores the advanced techniques. Specially, the reference frame memory compression and mode prediction in Geometry-motion-partition are researched deeply, and then the corresponding algorithms are proposed respectively.1、Reference frame memory compression(RFC):In order to reduce the memory cost and memory access bandwidth in video codec, an algorithm of2×2-block based reference frame memory compression is proposed. The algorithm assigns the limited bits among the pixels in a2×2-block, expoiting the high redundancy between them. In the mean time, the2×2unit can be reorganized to get the lowest memory access bandwidth according to the Align/Burst parameters of the employed memory. By this way, the reference frame memory and the memory access bandwidth is highly decreased when doing motion estimation or motion compensation in video terminal. Experimental results show that the compression algorithm can balance the coding performance and memory access bandwidth better compared with those previous methods based on4×4-block.2、Mode prediction in geometry-motion-partition(GEO):The algorithm of GEO mode prediction is proposed for further decreasing the coding bit rate when the GEO mode is adopted, which is based on the rigidity of objects’motion and the continuity of objects’boundary. Intra and inter predicted are both investigated. Then entropy coding is implemented to the residual data to reduce the bit rate. Experimental results show that the GEO mode prediction technology is available for coding chroma signal but limited prediction effect is conducted because of introducing more side information.
Keywords/Search Tags:Video Coding, HEVC, Reference frame memory compression, Geometry-motion-partition
PDF Full Text Request
Related items