Hyperspectral image dimension reduction method and device based on self-adaptive collaborative graph discriminant analysis

Abstract

The invention belongs to the field of remote sensing image processing, and discloses a hyperspectral image dimension reduction method and device based on self-adaptive collaborative graph discriminantanalysis. The method comprises the steps of selecting part of pixels from original hyperspectral data to serve as training samples; establishing a Tikhonov regular weight coefficient matrix partitioned by categories, and constructing a collaborative representation graph; and through generalized eigenvalue decomposition, obtaining an optimal projection matrix P under an optimization criterion, andprojecting a test sample to a low-dimensional space to realize dimension reduction of the hyperspectral data. According to the method, distance-weighted Tikhonov regularization is coupled with a representation method based on l2-norm minimization, data is projected to a low-dimensional popular space, and collaborative representation characteristics are obtained through l2-norm. In the process ofconstructing the graph, the internal relation between intra-class pixels is fully mined, and cooperative representation is adaptively adjusted through distance weighting measurement. In addition, thegraph weight matrix adopts a block diagonal structure design, so that the calculation cost is reduced, and the discrimination capability is further improved.

Description

technical field [0001] The invention belongs to the field of remote sensing image processing, and in particular relates to a hyperspectral image dimensionality reduction method and device based on adaptive collaborative graph discriminant analysis. Background technique [0002] Hyperspectral image target recognition technology has strong practicability and has been widely used in many fields such as public security, environmental monitoring, urban planning, geological survey, and medical diagnosis. Hyperspectral imagery (HSI) is an image obtained by a remote sensing system that records hundreds or even hundreds of continuous spectral bands. Due to the large number of bands and rich spectral information of hyperspectral images, it is possible to use them to accurately identify and classify ground objects. However, when processing hyperspectral images, the number of available training samples is often lower than the dimensionality (number of bands) between their spectra, whic...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a hyperspectral image dimensionality reduction method and device based on adaptive collaborative graph discriminant analysis to solve the problem that existing dimensionality reduction techniques affect the subsequent classification accuracy of hyperspectral images

Images