Adaptive Ratio for the Compressed Sensing of Multi-view Videos

Abstract

Most of the compressed sensing for multi-view videos always works on non-adaptive linear projections, and the ratio of compressive sampling is usually determined empirically. Consequently, the quality of reconstruction frames is always affected and limited. The paper is therefore proposed a method on the derivation of adaptive ratio for the compressive sensing of video frame. First of all, the block-based compressed sensing (BCS) with conventional selection for compressive measurements was reviewed. Then an estimation method for the sparsity of multi-view video frames was proposed with the two dimensional sparse transform (2DST). With an energy threshold determined beforehand, the DST coefficients were normalized and sorted by the descending order, and the sparsity of the multi-view video can be achieved by calculating the proportion of dominant coefficients. And finally, the simulation result shows that, the method can estimate the sparsity of a video frame effectively, and provides a basis for the selection of compressive measurement. The result also shows that the method can ensure the reconstruction quality of multi-view videos.