Font Size: a A A

Research On Entropy Coding In Video Coding

Posted on:2008-08-20Degree:MasterType:Thesis
Country:ChinaCandidate:R DaiFull Text:PDF
GTID:2178360272469110Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Video coding is the basis and key technology in multimedia communications, which will become an essential part in the communication systems for the next generation. Co-sponsored by National 863 Project"Digital Video and Audio Coding, Transmission, Testing and Application System (No. 2002AA119010)"and Hubei Science Foundation"Research of High Quality Image & Video Coding for Mobile Multimedia (NO. CGZ0223)", this thesis conducts an in-depth research on the entropy coding schemes in video coding. Meanwhile, our research group, the Multimedia Communications Research Group, has been a major member of China's Digital Audio and Video Coding Standard (AVS) Workgroup. By submitting technical proposals and taking our responsibilities in the workgroup, we help to push the drafting of the AVS Standard of China.Arithmetic coding is an efficient entropy coding scheme. The research on arithmetic coding in video coding has become a significant topic of research recently. A successful implementation of arithmetic coding in video coding standards is the Context-based Adaptive Binary Arithmetic Coding (CABAC) in H.264/AVC. The emerging AVS standard of China also includes context-based arithmetic coding as a normative part. There are three elementary steps in context-based binary arithmetic coding: binarization, context modeling, and binary arithmetic coding.This thesis proposes a novel arithmetic coding scheme for the number of nonzero coefficients in transform blocks. Firstly, the number of nonzero coefficients of each transform block is binarized according to the numbers of nonzero coefficients of its top and left neighbors. In order to simplify the binarization process, a novel mapping rule is designed to map a number of nonzero coefficient to a code number based on its unique probability distribution. Next, the binarized symbols are assigned probability models according to context information. Lastly, the binarized symbols are passed to the arithmetic coding engine and compressed. Experimental results show that compared to the conventional scheme, this scheme can reduce computational complexity with comparable coding efficiency.Another topic of the thesis is the design of"AVS P2 Golden C Decoder". One of the key recommended steps in developing an embedded video codec is to first develop a platform independent video codec implementation in what is called as"Golden C". The essence of our proposed decoder is that all the decoding operations are restricted in macroblocks. This feature enables parallel operations between macroblocks, reducing the spatial and temporal complexity for the whole decoding process. This thesis will introduce the design rules of the codec, the system design of the decoder, and the detailed design of the variable length decoding part. Since AVS P2 has been approved as a formal standard of China in March 2006, it can be predicted that the design of"AVS P2 Golden C Decoder"will bring in considerable applications in the future.
Keywords/Search Tags:Entropy coding, CABAC, adaptive binarization, mapping rules, context modeling, Golden C Decoder
PDF Full Text Request
Related items