Image parallel block compressive sensing scheme using DFT measurement matrix

Abstract Compressive sensing (CS)-based image coding has been widely studied in the field of processing. However, CS-based encoder a significant gap reconstruction performance compared with conventional compression methods. In order to improve quality encoder, we proposed an parallel block compressive (BCS) scheme, which is based on discrete Cosine transform (DCT) sparse basis matrix and partial Fourier (DFT) measurement matrix. BCS each column sampled by same DFT Due complex property matrix, compressed data complex. Then, real part imaginary resulting are quantized transformed into two bit streams, respectively. At stage, streams back signals using inverse quantization operation. The combined one signal, served as input CS reconstructed algorithm. theoretical analysis minimum Frobenius norm method demonstrates that outperforms other matrices. simulation results show better than matrices for BCS. Specifically, analyzed impact CS. experiment effect framework can nearly be ignored.