Unbalanced hyperspectral image classification method

Abstract

The invention discloses an unbalanced hyperspectral image classification method based on a depth generation spectrum-space classifier. The method mainly solves the problem that in the prior art, in the face of an unbalanced hyperspectral classification task, the classification precision of small classes is poor. According to the implementation scheme, a hyperspectral image is obtained, waveband selection is carried out, and the hyperspectral image is divided into a training sample and a test sample; constructing an unbalanced hyperspectral image classification network comprising a two-stage three-dimensional encoder, a three-dimensional decoder, a small-class up-sampling module and a classifier; training the classification network by using the training sample, setting an overall loss function, initializing network training parameters, and updating the classification network by using a gradient descent method until the maximum number of iterations is reached; and inputting a test sample into the trained classification network to obtain a classification result. According to the method, the classification precision of small categories in the hyperspectral image is improved, the robustness is enhanced, and the method can be used for mineral exploration, ecological monitoring, intelligent agriculture and medical diagnosis.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a non-equilibrium hyperspectral image classification method, which can be used for mineral exploration, ecological monitoring, smart agriculture and medical diagnosis. Background technique [0002] Hyperspectral images have high spectral resolution and can cover visible, near-infrared, and short-wave infrared wavelength ranges, usually with tens to hundreds of bands. The unique and fine spectral information of different substances makes it possible to distinguish land cover categories with only slight differences. Currently, it has been widely used in mineral exploration, ecological monitoring, smart agriculture, and medical diagnosis. Hyperspectral image classification is one of the important means of hyperspectral image interpretation. It is committed to assigning a specific label to each pixel in the image, so as to achieve pixel-level semantic analysis of...

AI Technical Summary

Problems solved by technology

However, the defect of this method is that the convex hull of the small category is far from the real data distribution, resulting in less useful information of the synthetic samples, which cannot solve the problem of insufficient small category samples, and the classification performance needs to be further improved.

Images