This paper introduces a framework for super-resolution of scalable videobased on compressive sensing and sparse representation of residual frames inreconnaissance and surveillance applications. We exploit efficient compressivesampling and sparse reconstruction algorithms to super-resolve the videosequence with respect to different compression rates. We use the sparsity ofresidual information in residual frames as the key point in devising ourframework. Moreover, a controlling factor as the compressibility threshold tocontrol the complexity-performance trade-off is defined. Numerical experimentsconfirm the efficiency of the proposed framework in terms of the compressionrate as well as the quality of reconstructed video sequence in terms of PSNRmeasure. The framework leads to a more efficient compression rate and highervideo quality compared to other state-of-the-art algorithms consideringperformance-complexity trade-offs.
展开▼
