An Adaptive Rate Allocation Method for Distributed Compressive Video Sensing

The rate-distortion performance of distributed compressive video sensing (DCVS) can be improved by adaptive block-based compressive sensing (ABCS). However, the original frames are not available in DCVS, ABCS cannot directly perform adaptive rate allocation according to the sparsity of blocks. To solve this problem, this paper proposes an inter-block correlation based adaptive rate allocation method for distributed compressive video sensing. Specifically, we define the inter-block correlation and calculate the inter-block correlation of each block from the initial measurement. By exploiting these inter-block correlations, more measurements are allocated to texture regions in the original frame, while fewer measurements are allocated to smooth regions. Meanwhile, we use the multihypothesis method to estimate the side information frame at the decoder. Experimental results indicate that the performance of the proposed scheme is effectively improved when compared with other adaptive sampling schemes.