Effective Image Block Compressed Sensing

In this paper, we propose an effective block compressed sensing algorithm using projected Land Weber based on bivariate shrinkage (BCS PL-BS) for natural images, and an improved noise variance estimation method is presented by using soft-thresholding bivariate shrinkage model for wavelet-based image denoising, which can more effectively remove noise and achieve better image reconstruction quality. Furthermore, the BCS PL-BS algorithm based on DPCM quantization is depthly studied. Experimental results demonstrate that the reconstruction performances of the proposed algorithm significantly outperform those of several state-of-the-art compressed sensing algorithms.