Rate-distoriton analysis for Compressive Sensing based coding

Abstract

Compressive Sensing (CS) is an emerging technology which samples a sparse signal at a rate corresponding to its actual information content rather than to its bandwidth. Different from traditional coding schemes in which distortion mainly comes from quantizer, distortion are related to quantization and compressive sampling in compressive sensing based coding schemes. Since the total coding bits are often constrained in the practical application, it is a great challenge to balance the number of measurements and quantization parameter to minimization the distortion. In this paper, a source rate and distortion model is proposed. The accuracy of the proposed R-D model is verified through experiments. Based on the R-D model, the optimal number of measurements and quantization step size are determined according to the rate-distortion criteria. Experimental results show that the proposed algorithm improves coding performances substantially.