Hyperspectral remote sensing image recognition method and device, electronic equipment and storage medium

Abstract

The invention discloses a hyperspectral remote sensing image recognition method and device, electronic equipment and a readable storage medium. The method comprises the steps that an image recognition model is trained in advance, a hyperspectral remote sensing image to be recognized is input into the image recognition model, and an image recognition result is obtained. A network input combination module of the image recognition model combines neighborhood features and spectral dimensions of the input image, and selects a network input image block; a multi-scale bi-pass convolution module performs fixed kernel convolution and multi-scale hole group convolution operation on the network input image block by using the first channel and the second channel respectively; a convolution feature mixing module is used for mixing multi-channel features in different convolution processes; a compression and expansion spatial spectrum attention mechanism module learns target features in a spectral dimension and a spatial dimension, adaptively adjusts a spatial spectrum feature map based on an attention weight, and transmits the finally obtained target spatial spectrum feature map to a discrimination convolution layer and a full connection layer for dimension reduction operation, thereby effectively improving the recognition precision of a hyperspectral image.

Description

technical field [0001] The present application relates to the technical field of image processing, in particular to a hyperspectral remote sensing image recognition method, device, electronic equipment and readable storage medium. Background technique [0002] With the development of hyperspectral sensors, the spectral information of remote sensing images is becoming more and more abundant. With many fingerprint features, remote sensing technology has entered a new stage. Hyperspectral imagery has a spectral sampling interval of up to nanometers, and the available pixels have spectral information of hundreds of bands. Among them, in the field of ground object classification, hyperspectral image classification is to assign each pixel a unique category label according to the spectral and spatial characteristics contained in the image, which is currently the most active research field in hyperspectral remote sensing technology, and has attracted extensive attention in the fiel...

AI Technical Summary

Problems solved by technology

Convolutional networks are generally based on patches. In hyperspectral remote sensing image classification, the setting of the block size will lead to distortion of the classification results.

Due to the poor ability to perceive details in space, the excessive increase of network depth will bring problems such as gradient disappearance to the network model, and at the same time, due to the complex network structure and limited model training samples, the model training is insufficient and the recognition accuracy is low.

Images