Context-based Four-dimensional Matrix Video Coding

With the rapid development of digital communication technology, the mature of image compression techniques and the publication of international standards, video communication has came into human's daily life and works. A large quantity of image and video information need be transmitted and stored, which brings new requirements and problems to transmission schemes and storage capacity. It is important to improve compression and decompression performance of digital video under the finite bandwidth of network.Pervasive, seamless, high-quality digital video has been the goal of companies, researchers and standards groups over the last two decades. This has led to the development of a number of key international standards for image and video compression, including the MPEG-1, MPEG-2, MPEG-4 and H.26x series of standards. Most of these standards are based on the method of inter-frame motion compensation and two-dimensional discrete cosine transform (2D-DCT) and encode and describe the color video in YCbCr 4:2:0 format, which want to take advantage of human visual system (HVS) to save bit expense by decreasing the resolution of two color difference components. Then, the gray image encoding method is separately applied to each component. As we all known, the R, G, B components of a frame describes the same texture and margin of the image. Each one of them can almost describe the information of the whole image except color. If encoding them separately, it will not take enough advantage of the correlation between the components.This thesis is part of the project of National Science Foundation of China"Study on Color Video Coding Based on Four-dimensional Matrix". A multi-matrix theory model has been established in the project. And the color video is presented into a four dimensional matrix model and the four-dimensional discrete cosine transform (4D-MDCT) is applied to reduce the correlation between color components and adjacent pixels. However, there still exists correlation between coefficients of sub-matrices after 4D-MDCT. This thesis proposes a four dimensional matrix video coding method based on context to get rid of the correlation between coefficient sub-matrices further. It includes three parts as follows:Firstly, this thesis presents a four-dimensional matrix video encoder and decoder model, which contains components of prediction, transform and entropy coding.