Research Of Distributed Compressive Video Sensing

The theory of compressed sensing has a wide application prospect in other fields such as image video and other signal processing.Under the distributed video compression,this paper further studies how to reduce the number of samples and improve the quality of the side information,and combine these two methods to design the video codec.First of all,in order to reduce the sampling of redundant information and the data processing during coding,an adaptive sampling rate allocation method based on image entropy is proposed.This method samples the key frames at a fixed sampling rate,and the CS frames use the information entropy of the block as the weight of the assigned sample rate.The simulation results show that the data processed by this method is greatly reduced.Secondly,when the edge information is generated,this paper proposes an optimized edge information generation algorithm based on the spatio-temporal correlation of video sequences,aiming at the problem that the pixels in the same image have inconsistent motion vectors.The algorithm uses a new way of variable block-based distributed motion estimation for different motion regions.The proposed algorithm improves the video sequence to a certain degree.Finally,the two algorithms proposed are applied to video coding and decoding.The adaptive sampling theory based on information entropy is used to sample and encode the video sequence at the coding end.The edge information based on the spatial-temporal correlation property is generated at the decoding end.The results showing that the codec designed in this paper can significantly improve the reconstruction efficiency when the sample size are small.