Fast Scalable Video Coding Method Based On Compressed Sensing

With the wide application of wireless internet,the demand for compression processing of HD video is increasing,the traditional video coding can only encode a single character stream,but it can not fully adapt to the various user needs.Therefore,it is the key point of video coding research to adapt to the changeable situation by researching scalable video coding.However,the coding efficiency would have a certain degree of decline due to reasons such as fine quantization in the enhancement layer of scalable video coding,so by using the sparsity of compressed sensing to solve this problem has greatly practical significance.In this thesis,a universal scalable coding framework based on compressed sensing is proposed.Then,by using H.264 SVC as a concrete object,three kinds of scalable fast algorithms are proposed for different conditions of video coding on the base of this framework.The main tasks are as follows:(1)A fast quality scalable video coding algorithm based on compressed sensing is proposed.The main idea of the algorithm is that the base layer of the scalable video coding is kept unchanged and quickly obtaining the encoding scheme of the enhancement layer by using the interlayer correlation,and then the part in the enhancement layer which needs to be fine quantified is combined with the compressed sensing theory.Video would be coded sparsely and transimitted after setting up flag bit and adding 0 behind measurement value,and the identification judgment of the flag bit is used to reconstruct the compressed sensing in the decoding end.(2)A fast spatial scalable video coding algorithm based on compressed sensing is proposed.This algorithm makes full use of the interlayer correlation between the base layer and the enhancement layer,and combines the sparse representation theory with the uncoded subblock in enhancement layer of the spatial scalable video encoding.The main idea of the algorithm is.to keep base layer immutable and performing sparse coding for enhancement layers,which satisfy the sparse representation after the fast mode selection,after setting up mark and add 0 behind measurement value to reduce coding rate of enhancement layer.(3)A fast mixed scalable video coding algorithm based on compressed sensing is proposed.This algorithm is based on the idea of hierarchical frame structure,and combines the compressed sensing theory with the quality and temporal scalable video coding.The main idea of the algorithm is.that the frames in base layer keep its original coding method and frames selected by fast hierarchical prediction algorithm.in enhancement layer are coded as same as previous twoalgorithms.At the decoding end,it is necessary to determine whether the compressed sensing reconstruction is required by judging the status of the flag bit.