Method and system for extracting ground object spatial spectral features of hyperspectral remote sensing image

Abstract

The invention belongs to the technical field of image processing, and discloses a method and system for extracting ground object spatial spectral features of a hyperspectral remote sensing image. Themethod comprises the steps of training and extracting spectral features through an auxiliary classifier generative adversarial network; performing band selection, and extracting spatial texture features with rotation invariance from a selected band; and forming spatial spectral features of the ground object through splicing the spectral features and the spatial texture features. Meanwhile, the invention discloses a hyperspectral remote sensing image classification system which adopts the ground object spatial spectral features is based on a convolutional neural network. The method and system verify that the ground object spatial spectral feature extraction technology disclosed by the invention not only can better characterize ground object information, but also can obtain higher classification accuracy with fewer labeled data sets.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a method and system for extracting spatial spectrum features of ground objects in hyperspectral remote sensing images. Specifically, it relates to an ACGAN-based method and system for extracting spatial spectral features of ground objects in hyperspectral remote sensing images. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] The theoretical basis of hyperspectral remote sensing image surface object classification is that different surface objects have different absorption and reflectance of spectral signals in different bands, and have different surface object spectral characteristic curves. According to the known typical spectral curves of surface objects, the pixels in hyperspectral remote sensing images can be classified into a certain object category. The application fields of h...

AI Technical Summary

Problems solved by technology

These methods usually require more labeled samples, are not suitable for multi-category and small sample situations, and have poor applicability

Research by Christian Szegedy and others found that deep learning models including CNN are extremely vulnerable to adversarial samples, and adversarial samples have become a blind spot for training algorithms

Images