A two-stage multi-hypothesis reconstruction scheme in compressed video sensing

Abstract

Existing multi-hypothesis (MH) prediction algorithms in compressed video sensing (CVS) are all deployed in measurement domain, which restricts the flexibility of block partitioning in the reconstruction process and decreases the reconstruction accuracy. To address this issue, this paper proposes a two-stage multi-hypothesis reconstruction (2sMHR) scheme which deploys the MH prediction in measurement domain and pixel domain successively. Two implementation schemes, GOP-wise and frame-wise scheme, are developed for the 2sMHR. Furthermore, a new weighted metric combining the Euclidean distance and correlation coefficient is designed for the Tikhonov-regularized MH prediction model. Simulation results show that the proposed two-stage MH reconstruction scheme obtains higher reconstruction accuracy than the state-of-the-art CVS prediction methods.