Distributed compressed video sensing based on the optimization of hypothesis set update technique

Abstract

Compressed Sensing (CS) breakthroughs the limitation of Nyquist sampling rate and realizes the sampling and compression of data simultaneous. Hence, it is widely used in image processing and video compression. However, it remains a challenge to obtain the high quality reconstructed image and video. To this end, we focus on the reconstruction algorithm of CS and the melioration of the existing distributed compressed video sensing (DCVS). In the perspectives of hypothesis set design and reference frames selection, we give detailed analyses for existing schemes and propose hypothesis set updating (HSU) and dynamic reference frames selection (DRS) algorithms to polish up the reconstruction performance. Then the superiorities in performance for these schemes are illustrated. Finally, the simulation results indicate that at a low sampling rate, the block based DCVS with the proposed HSU and DRS (HD-BDCVS) ameliorates the quality of non-key frames and key frames simultaneously without increasing the complexity of the coding.