Method, apparatus and device for determining compressed sensing measurement matrix

Abstract

The invention discloses a method for determining a compressed sensing measurement matrix, and the method comprises the steps: constructing an N*N-order orthogonal square matrix based on a sparse matrix; extracting M row vectors from the orthogonal square matrix to generate an M*N-order initial measurement matrix; then, using the cross correlation coefficient between the column vectors as an evaluation basis for verifying the randomness of the measurement matrix; performing iterative operation of updating element positions on the basis of the initial measurement matrix; taking the measurement matrix with the minimum cross correlation coefficient in each iterative operation as the optimized measurement matrix, wherein compared with the initial measurement matrix, the optimized measurement matrix has higher randomness, improves the reconstruction precision, and is beneficial to meeting the acquisition requirements of more original signals. The invention further discloses an apparatus anda device for determining a compressed sensing measurement matrix and a computer readable storage medium, which have the above beneficial effects.

Description

technical field [0001] The present invention relates to the technical field of signal processing, in particular to a method for determining a compressed sensing measurement matrix, a determining device, a determining device, and a computer-readable storage medium. Background technique [0002] Compressed sensing technology breaks through the limitations of the traditional Nyquist sampling theorem, can greatly reduce the sampling rate of signals, realizes data acquisition and compression at the same time, and avoids the transmission of a large number of redundant signals. The natural signal undergoes sparse transformation to obtain a sparse matrix containing K non-zero values, and then through the measurement matrix processing, the signal is projected into a low-dimensional space, and the signal is reconstructed in the low-dimensional space to obtain the signal acquisition result. [0003] One of the difficulties in compressed sensing technology is how to construct a stable m...

AI Technical Summary

Problems solved by technology

However, this method of constructing the measurement matrix greatly reduces randomness and can only satisfy the RIP characteristics of a certain order. The probability of successful reconstruction is highly dependent on which row vectors are extracted. If the cross-correlation between the extracted row vectors is compared Large, there is a certain possibility of reconstruction failure, which will further lead to a smaller value of the sparsity K that can be restored, so it is only suitable for sparse signals in the time domain, and cannot satisfy most original signals such as natural images.

Images