Hyperspectral image classification method based on feature random sampling integrated extreme learning machine

Abstract

The invention discloses a hyperspectral image classification method based on feature random sampling and integrated extreme learning machine. The existing hyperspectral image has the characteristics of map-spectrum integration, high spectral resolution, wide spectral range, and strong spectral correlation. This invention aims at the problem that the hyperspectral image classification method is complicated and difficult to implement in real time, and proposes a method based on neighborhood spectral information extraction. Space spectrum feature, in order to reduce the complexity of algorithm design, considering the correlation between the adjacent bands of space spectrum feature, the original space spectrum feature vector is averagely grouped, and then several features are randomly selected from each interval to combine, using The fast learning ability of the extreme learning machine is used to train the weak classifier, and finally the hyperspectral image classification is realized through the voting method. Tests show that this method does not require a complicated optimization process, has fast training speed and high classification accuracy, and can meet the requirements of hyperspectral image classification accuracy and real-time performance.

Description

technical field [0001] The invention belongs to the technical field of hyperspectral remote sensing image classification, and relates to a hyperspectral image classification method based on feature random sampling and integrated extreme learning machine. Background technique [0002] Hyperspectral remote sensing combines imaging technology with subdivision spectrum technology. While imaging the spatial characteristics of the target, each spatial pixel is dispersed to form dozens or even hundreds of narrow bands for continuous spectral coverage. The acquired hyperspectral image contains rich triple information of radiation, space and spectrum, and is a comprehensive carrier of various information. [0003] Hyperspectral imagery has the characteristics of map-spectrum integration, high spectral resolution, wide spectral range, and strong spectral correlation, making it play an important role in target reconnaissance, geological exploration, marine environmental monitoring, agr...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem that the hyperspectral image classification method is complex and difficult to real-time, and propose a hyperspectral image classification method based on feature vector average grouping, random sampling, and integrated extreme learning machine. This method can quickly Complete hyperspectral image classification accurately, which can meet the requirements of hyperspectral image classification accuracy

Images