Hyperspectral target detection method based on unmixing pretreatment

Abstract

The invention provides a hyperspectral target detection method based on unmixing pretreatment. The method comprises the following steps: 1) through a detected hyperspectral image, obtaining a target spectrum t needing to be detected, and carrying out unitization processing on the hyperspectral image and the target spectrum; 2) carrying out end member extraction on the hyperspectral image to obtainan end member set of the image; 3) carrying out spectrum included angle calculation on the obtained end member set obtained in the step 2) and the target spectrum t to obtain a target end member t <~>most similar to the target spectrum in the end member set; if the target end member cannot be found in the set threshold value, projecting the hyperspectral image to an orthogonal subspace of principal components thereof; repeating the steps 2) and 3) until matching the target end member t <~>; 4) carrying out abundance inversion on the target end member obtained in the step 3) to obtain an abundance image of the target end member; 5) carrying out processing on the abundance image obtained in the step 4) to obtain optimal segmentation threshold of the abundance image; and 6) carrying out segmentation on the abundance image according to the threshold obtained in the step 5), wherein the white region in the image obtained after segmentation represents a target region and the black region represents a background region.

Description

technical field [0001] The invention belongs to the technical field of hyperspectral target detection, and in particular relates to a hyperspectral target detection method based on unmixing preprocessing. Background technique [0002] Hyperspectral target detection technology is an important direction in the application of hyperspectral remote sensing technology, which has been widely used in military and civilian fields. A hyperspectral image has three dimensions, including two spatial dimensions and one spectral dimension. In the hyperspectral image, each pixel has hundreds of continuous spectral bands, and the width of these bands is usually about 10nm. The pixels of the same band in the hyperspectral image form a 2D image, so the hyperspectral image has "Integrated graphs" feature. [0003] In the past few decades, several hyperspectral object detection algorithms have been proposed. These include Spectral Angle Mapping (SAM), Constrained Energy Minimization (CEM), Ad...

AI Technical Summary

Problems solved by technology

[0005] The present invention proposes a hyperspectral target detection method based on unmixing preprocessing on the basis of the NCLS algorithm, improves the existing NCLS algorithm, and solves the problem of poor detection effect if the target cannot be successfully extracted as an end member in the existing NCLS algorithm This problem, and using the particle swarm optimization algorithm to optimize the threshold of the abundance inversion

Images