Adaptive Coding for Compressed Video Sensing

In this paper, we propose an Adaptive Coding based Compressed Video Sensing (ACCS) scheme for Distributed Video Coding. Our scheme mimics the traditional video coding method and performs the mode decision both at the encoder and the decoder. At the encoder, the ACCS divides the frame blocks into three categories: the SKIP mode, INTER mode and COMBINED mode according to the residual of the blocks, and the adaptive sampling rate is chosen for these modes. At the decoder, we adopt different decoding methods for different modes. For the COMBINED mode, we apply adaptive decoding scheme by exploiting the intra-frame and inter-frame sparsity. Experimental results show that the proposed algorithm outperforms existing state-of-the-art video CS approaches at a very low sampling rate.