Image compression and reconstruction in compressive sensing paradigm

Abstract

Compressive sensing (CS) is a new branch of research with applications in signal processing, medical imaging, seismology, communications, and a variety of other fields. It assures successful data compression and faithful reconstruction by considering a smaller number of linear measurements compared to its dimensions. In this paper, we have shown CS paradigm for image compression and reconstruction. We have considered the Basis Pursuit (BP), Lp – Reweighted (Least Squares Method), Orthogonal Matching Pursuit (OMP) and Fusion of OMP & BP algorithms to obtain the compressive measurements and faithful reconstruction. The results are analyzed by varying sparsity level and Compression Ratio (CR) and then calculating the Peak Signal to Noise Ratio (PSNR) value. The obtained results show that OMP performs better for standard test images & satellite images and Fusion of OMP & BP performs better for biomedical images.