A multihypothesis-based residual reconstruction scheme in compressed video sensing

A multihypothesis-based residual reconstruction scheme (MHRR) is presented in compressed video sensing (CVS). The residual is first predicted by a novel multihypothesis (MH) prediction method and the second prediction residual then is reconstructed independently. In the proposed MHRR, a method of generating hypothesis blocks in residual domain is offered, and the hypothesis blocks are obtained by pixel-domain ME technique and the linear prediction weights are calculated in measurement-domain, which can combine the advantages of pixel-domain ME technique and measurement domain MH prediction. Simulation results show that the proposed MHRR can achieve higher reconstruction performance than SPL (Smoothed Projected Landweber) for residual reconstruction.