Research Of Video Compression Based On Compressed Sensing Over Wireless Video Multicast

With rapid development of wireless network and mutimedia rechniques, vedio muticast is possible in wireless environment. How to provide good services to ensure different users is concentrated . These channels as well as users differ with bit rates, packet loss rate.But they all need high-quality video.This article tries to solve the problem above by taking advantage of the latest theory of compressive sensing . It deals with signals by measurement which leads to the unstructured compressed data. We try to design an appropriate video codec for wireless multicast environment by taking advantage of the features mentioned above.Firstly, parts of compressed sensing data were used for vedio frame reconstruction,the result showed that certain quality vedio can still be reconstructed.Experiments showed that the reconstruct video quality was proportional to the amount of compressed sensing data .Then The characteristics above were extended to the wireless multicast,experiments also showed that reconstruct video quality was proportional to bit error rate in wireless environmentTo further enhance video compression, three different scenarios were proposed.The first was a secondary compressed sensing which suit for security vedio. In this case,difference between the non-critical frame and key frames experienced compressed sensing again . Then according to exsiting motion estimation framework as well as high sparsity characteristics of the residual,key frames together with residuals are compressed respectively.Finally ,by taking advantage of correlation between frames,redundant dictionaries was introduced, key frames and non-key frames were compressed respectively at frame-based level and block -based level.Experimental results showed that compressed sensing based video codec met the needs of the multiple receiving in wireless multicast environment,that is to say, A single stream can met the needs of mulitiple channels.