Measurement Matrix of Compressive Sensing Based on Gram-Schmidt Orthogonalization

Abstract

Measurement matrix plays an important part in sampling data and reconstructing signal in Compressive Sensing (CS). In this paper, the common measurement matrices and the relationship between measurement number of measurement matrix and signal sparsity are researched. The performance among the common measurement matrices is compared. In order to obtain a better reconstruction result, an improved method based on Gram-Schmidt orthogonalization of row vectors for matrix is proposed. The experiments show that the improved measurement matrix is better than the original measurement matrix when used to reconstruct signal.