Study On Key Technoloyg Based On Wyner-Ziv Video Coding

With the development of wireless network, more and more mobile video terminals are used in multimedia communications. Since these wireless devices are always limited in computation capability and energy resources, the encoder should be low-complexity and error resilience with high compression efficiency. The traditional video compression standards such as MPEG and H.26x could not be used in error-prone and limited resources wireless network. To tackle the problem, the scholars home and abroad proposed the novel video compression method:Wyner-ziv video coding.This thesis mainly researches on it and the major contributions are as follows:1.On the basis of studying structures of the existing Wyner-ziv video coding, the thesis analyzes its advantages and problems, which includes balancing the coding complexity and compression efficiency at encoder, probability statistics model between current decoded frame and side information, rate allocation between Wyner-ziv frame and key frame and how to generate accurate side information based on motion estimation and reconstruct frames jointly.2.This thesis proposes a hierarchical variable size block matching (HVSBM)algorithm to do motion estimation and generate side information at decoder.This scheme is carried on the structure of low bit rate wavelet-based Wyner-Ziv video coding. In this scheme, HVSBM motion estimation is used to scearch process at the decoder,which makes the magnitude of sum of absolute difference of motion vector between the former level and the later level assumed split as a bechmark to decide the later level wether to split or not,and lastly generates a tree structure of three-level motion vector blocks.The fomer and later key frames do forward and backward HVSBM motion estimation respectively and calculate the interploation.Then this interploation is used to predict Wyner-Ziv frame as side information.The experimental result shows that at the cost of increasing (rather small) encoder complexity, this proposed scheme achieves better performance generally compared with the fixed block size method. At the same output bit rate, the PSNR of codec increases about 0.15dB averagely compared with the original algorithm and achieves better subjective improvement. 3.This thesis implements DCT domain-based Wyner-Ziv video coding.On the basis of deeply study on every module of Wyner-Ziv video coding, Wyner-Ziv video coding is emulated,which includes rate control with low complexity and non-feedback,side information generating at the decoder and LDPC decode etc. Then, the performances of Wyner-Ziv video coding is made and analyzes in details.