Hyperspectral image classification method based on global attention residual network

Abstract

The invention discloses a hyperspectral image classification method based on a global attention residual network, and the method comprises: constructing an overall network, which comprises a multi-scale feature extraction network, a global attention module and an improved residual network module; performing multi-scale feature extraction to extract hierarchical features of the hyperspectral image; constructing a spatial and spectral dependency relationship of global pixel points by the global attention module through combination of the spatial attention module and the spectral attention module; fusing the improved residual error network module and the global attention module to form a novel global attention residual error network; and sending an output result into a classifier through global pooling for final classification, and outputting a result. According to the method, rich spatial-spectral features are obtained at the same time by introducing a multi-scale receptive field and a global attention module, and an improved residual network is added to relieve the gradient disappearance problem and accelerate network convergence, so that the classification precision is improved, and a good and stable classification effect is ensured.

Description

Technical field[0001]In the field of high spectral remote sensing image processing, the present invention, in particular, there is a high spectral image classification method based on a global attention residual network.Background technique[0002]High-spectral images (HSI HYPERSPECTRAL IMAGES have been significantly applied in various remote sensing, agricultural monitoring, marine safety, etc. in recent years, high-spectable images are different from conventional two-dimensional digital images, which is a three-dimensional cube data. , Consist of two-dimensional digital images and one-dimensional spectral dimension. The spectral wavelength includes rich geographic feature information, and therefore, feature selection and feature extraction is especially important for high spectrum pixel points. Previously, for high-spectral images, common classification methods include: K nearest neighbor (K-NN, K NEAREST neighbour) [6], extreme learning machine (ELM, EXTREMELEARNING), and support v...

AI Technical Summary

Problems solved by technology

[0005] Purpose of the invention: In order to overcome the problems of insufficient utilization of image features, limited range of receptive field and difficulty in convergence of network structure redundancy in existing hyperspectral remote sensing image classification algorithms in the prior art, provide a residual network based on global attention Hypers...

Images