Hyperspectral image classification method based on cross-grouping spatial-spectral feature enhancement network

Abstract

The invention discloses a hyperspectral image classification method based on a cross-grouping spatial-spectral feature enhancement network. The hyperspectral image classification method comprises thefollowing steps: 1) cross-grouping spectral features; 2) extracting multichannel grouping spectral characteristics; 3) performing spatial feature cross grouping; 4) extracting grouping space features;5) performing spectral space channel information interaction; and 6) performing hyperspectral image pixel classification. According to the method, spectral spatial feature information is utilized toperform cross grouping and feature extraction operation on spectral features and spatial features respectively, so that the correlation between adjacent spectra can be effectively weakened; channel self-attention and pixel position self-attention operations are adopted to enhance features obtained by cross grouping, information interaction and fusion are carried out on spatial features and spectral features, the fused features are used for classification, and the network classification performance can be improved.

Description

technical field [0001] The invention relates to the technical field of intelligent image processing, in particular to a hyperspectral image classification method based on a cross-group spectral-spatial feature enhancement and fusion network. Background technique [0002] Hyperspectral remote sensing image (Hyperspectral image, referred to as HSI) has the characteristics of both spatial information and spectral information, which makes it of great application value in research fields such as object classification, target segmentation and recognition, and dynamic target tracking. At present, it has been widely used in agricultural detection, mineral exploration, safety monitoring and environmental science and other fields. The classification research of hyperspectral remote sensing images is one of the important means of hyperspectral image information extraction and the basis of hyperspectral research technology. By dividing the difference of each spectral band of different g...

AI Technical Summary

Problems solved by technology

First of all, due to the large number of spectral bands and large amount of data in hyperspectral images, directly using the original hyperspectral data blocks as the input of the model is prone to the curse of dimensionality; second, the computational complexity of 3D CNN is higher than that of 2D CNN. The model is prone to overfitting problems and cannot extract deeper features; finally, these methods do not take into account the strong correlation between adjacent spectra of hyperspectral images when extracting spectral features, ignoring the processing of spectral dimensions

Images