Two-dimensional orthogonal matching pursuit optimization algorithm based on singular value decomposition

Abstract

The invention discloses a two-dimensional orthogonal matching pursuit optimization algorithm based on singular value decomposition. The algorithm comprises acquiring a row measurement matrix, a columnmeasurement matrix, measurement values, a sparse base and signal sparseness. SVD decomposition is executed on two measurement matrixes, the measurement matrixes and the measurement values are updated, a residual, an index set and an optimized sensing matrix are initialized, an index is found, and new approximation of a signal is computed; the residual is further updated, continuous iterations areperformed, and at last the estimated value of the signal and the index set are output. According to the algorithm provided by the invention, separation of the measurement matrix in front-end information collection and rebuilt matrix in rear-end rebuilding are achieved, and thus the algorithm is applicable to a common separable linear system. The two measurement matrixes are subjected to SVD decomposition, so that the two optimized rebuilt matrixes are acquired, correlation between the measurement values is effectively eliminated, and the rebuilding signal to noise ratio and the robustness ofthe algorithm are significantly improved. Furthermore, the separable operator is used in the design of the measurement matrixes, and thus the algorithm can be applied to the building process of the large-scale image.

Description

technical field [0001] The invention relates to the technical field of image processing and compressed sensing imaging systems, in particular to a two-dimensional orthogonal matching pursuit optimization algorithm based on singular value decomposition. Background technique [0002] Compressed Sensing (CS) is a new framework in information acquisition and processing. Compared with the general framework that first collects as much data as possible and then discards redundant data through digital compression technology, CS tries to reduce the Collection of redundant data in the information collection step. Data compression is performed while acquiring data, which not only greatly reduces the amount of information collection, shortens the time of information collection, but also saves storage space. [0003] CS theory includes three key technologies: sparse representation of signals, incoherent measurement, and reconstruction algorithms. The fast and efficient reconstruction a...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a two-dimensional orthogonal matching pursuit optimization algorithm based on singular value decomposition to solve the problems that the 2DOMP algorithm does not have optimal performance.

Images