A hyperspectral image sparse unmixing method based on mfocuss and low-rank representation

Abstract

The invention discloses a hyperspectral image sparse unmixing method based on MFOCUSS and low-rank representation. The specific steps of the invention include: reading original hyperspectral data and known spectral library data; constructing a sparse solution based on MFOCUSS and low-rank representation The objective function of the mixed model, and the hyperspectral data and the known spectral library are used as the input data and dictionary of the objective function, and the abundance matrix of the entire spectral library is obtained by solving the objective function of MFOCUSS and the low-rank representation model, and the spectral library is eliminated. The spectrum of the non-true endmembers, the spectral library after removing the non-real endmembers, iterates repeatedly to finally obtain the real endmember matrix and the corresponding abundance matrix; this method avoids directly extracting the endmembers from the original hyperspectral data, and eliminates The non-true endmembers and the spectral library were updated, which reduced the adverse effect of the autocorrelation of the endmember spectra in the spectral library on the hyperspectral unmixing effect, and improved the accuracy of the abundance estimation.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to a hyperspectral image sparse unmixing method based on Multiple FOCal Underdetermined System Solver (MFOCUSS) and low-rank representation in the technical field of sparse unmixing. Background technique [0002] Hyperspectral remote sensing images are three-dimensional image data composed of dozens or even hundreds of continuous band images, which can organically combine images that determine the spatial characteristics of ground objects with spectral information that determines the properties of ground objects, and can qualitatively and quantitatively analyze the measured objects. Physical analysis and identification, so it is widely used in the fields of land resource utilization, geological exploration, environmental monitoring, marine resource survey, artificial target identification and military detection. However, due to the limitation of optical devices, the pixels in...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to overcome the deficiencies of the above-mentioned prior art, propose a hyperspectral image sparse unmixing method based on MFOCUSS and low-rank representation, reduce the unfavorable effect of the endmember spectral curve in the spectral library on the hyperspectral unmixing effect To improve the accuracy of sparse unmixing of hyperspectral images and overcome the problem of low efficiency of sparse unmixing of hyperspectral images

Images