A CS-based adaptive sampling rate surveillance video codec framework

Compressed sensing (CS) theory has indicated that sparse signals can be probably recovered from far less sampled data than the Nyquist Sampling theorem required. The sparsity of the video frame varies over time, so the quality of the reconstruction from the fixed sampling rate framework will fluctuate. To solve this problem, we have proposed a new CS-Based video codec framework with the adaptive sampling rate, which is predicted by the test of intra-frame sparsity and inter-frame sparsity. The simulation and analysis have proven that the proposed framework has a better performance on the reconstruction than the fixed sampling rate framework.