Low Complexity Distributed Video Coding Using Compressed Sensing

Compressive sensing (CS) is an efficient method toreconstruct sparse images with under-sampled data. In thismethod sensing and coding steps integrated to a one-step, lowcomplexitymeasurement acquisition system. In this paper, weuse a Non-linear Conjugate Gradient (NLCG) algorithm tosignificantly increase the quality of reconstructed frames of videosequences. Our proposed framework divides sequence of a videoto several groups of pictures (GOPs), where each GOP consistingof one key frame followed by two non-key frames. CS is thenapplied on each key and non-key frame with different samplingrates. For reconstruction final frames, NLCG algorithm wasperformed on each key frame with acceptable fidelity. To achievedesired quality on low-rate sampled non-key frames, NLCGmodified using side information (SI) obtained from last twosuccessive reconstructed key frames. Based on some performancemeasures such as SNR, PSNR, SSIM and RSE, ourimplementation results indicate that employing NLCG withGaussian sampling matrix outperforms other methods in qualitymeasures.

Keywords:

compressed sensing (CS) , distributed video coding DVC) , sparse reconstruction , nonlinear conjugate gradient (NLCG)