High-spectrum image end member selection method based on linear least-squares support vector machine

Abstract

The invention provides a high-spectrum image end member selection method based on an LLSSVM (Linear Least-Squares Support Vector Machine), comprising the following steps of: 1. selecting N pixel points as initial end members; 2. using the end element in the i position of the present selected end members as '1' class and the rest of N-1 end members as '0' class, executing i=1 for the first time, and establishing corresponding LLSSVM discrimination functions, namely distance measuring and calculating functions; 3. sequentially calculating the distance of each pixel, if the absolute distance of one pixel is greater than 1, substituting the end member in the i position for the pixel, setting the i to be equal to 1 and then switching to the step 2; 4. i=i+1, if i is greater than N, switching to step 5, and otherwise, switching to the step 2; and 5. finishing when the current end member is the final selected end member. The high-spectrum image end member selection method is accomplished by adopting the LLSSVM as a main tool and has the advantages of no need of dimension-reduction pretreatment and low complexity.

Description

technical field The invention relates to an endmember selection method of a hyperspectral image, in particular to an endmember selection method based on a support vector machine (SVM), and belongs to the technical field of remote sensing information processing. Background technique The spatial resolution of hyperspectral images is generally low, which leads to the widespread existence of mixed pixels, that is, a pixel may be a mixture of several categories. The technique of analyzing the proportion of each type of component in the mixed pixel is called spectral unmixing, which is one of the most basic and important contents of hyperspectral data analysis. The necessary prerequisite for the implementation of spectral unmixing is to know which object categories are included in the hyperspectral data. Under this requirement, the technique of extracting representative pure spectra of each category is called spectral endmember selection, or endmember selection for short. In the ...

AI Technical Summary

Problems solved by technology

CHOWDHURY A. et al. used sequential selection instead of joint selection. This sequential selection is far from the basic characteristics of the N-FINDR algorithm. Once a pixel is selected, it cannot be updated. Dependencies are also not maximally satisfied

The method proposed by TAO XUETAO et al. can be performed directly on the original data space without dimensionality reduction preprocessing, so the selected spectral endmembers are more reasonable, and theoretically break through the traditional mode of N-FINDR algorithm that requires dimensionality reduction processing. But this method also belongs to sequential selection

On the other hand, the endmember selection method based on convex geometry analysis is susceptible to the influence of outlier points, and there are a large number of outlier points in hyperspectral images, and the existing literature has not proposed a corresponding solution

Images