Hyperspectral image classification method and device and electronic equipment

Abstract

The invention provides a hyperspectral image classification method and device and electronic equipment, and the method comprises the steps: inputting a target hyperspectral image into a pre-trained feature extraction network model, and outputting the image features of the target hyperspectral image; determining an initial category of each pixel point in the target hyperspectral image based on the image features; inputting the image features into a pre-trained image reconstruction network model, and outputting a reconstructed image of the image features; based on the reconstructed image and the target hyperspectral image, determining a reconstruction loss value of each pixel point in the reconstructed image; and determining a final category of the pixel points according to the reconstruction loss value and the initial category. In the mode, the extracted image features are reconstructed to restore the features of the target hyperspectral image as far as possible, the reconstructed image is obtained, the pixel points of the unknown class can be determined according to the reconstructed image and the target hyperspectral image, and the accuracy of the classification result is improved.

Description

technical field [0001] The invention relates to the technical field of classification and recognition of hyperspectral images, in particular to a classification method, device and electronic equipment for hyperspectral images. Background technique [0002] A hyperspectral image is a three-dimensional light cube with rich spectral information and spatial information, which can be regarded as a three-dimensional image with one-dimensional spectral information in addition to ordinary two-dimensional images; the classification of hyperspectral images refers to the Extract spectral information or spatial information to classify each pixel in the image. In related technologies, it is usually based on the fact that the pixels are complete known categories, and each pixel can be classified using the existing label value to mark and classify the hyperspectral image. However, due to the large coverage area of ​​the hyperspectral image , usually the existing label values ​​cannot cove...

AI Technical Summary

Problems solved by technology

In related technologies, it is usually based on the fact that the pixels are complete known categories, and each pixel can be classified using the existing label value to mark and classify the hyperspectral image. However, due to the large coverage area of ​​the hyperspectral image , usually the existing tag values ​​cannot cover all the categories in the image, and there must be unmarked categories (unknown categories)

In existing classification methods, unknown classes are usually assigned to currently known class label values, resulting in larger estimated areas of known classes and reducing the accuracy of classification results

Images