Hyperspectral Anomaly Detection Method Based on Adversarial Autoencoder Network

Abstract

The invention discloses a hyperspectral image anomaly detection method based on an adversarial self-encoding network, which mainly solves the problems of complicated calculation and low detection accuracy in the prior art. The implementation plan is: 1) Use the pixel update method to make a hyperspectral image training data set; 2) Input the training data set into the GAN training, and extract the spectral features of the training data set; 3) Process the spectrum using band fusion and attribute filtering methods feature, to obtain the spatial feature of the training data set; 4) use the spatial feature to enhance the abnormal target in the original hyperspectral image; 5) use the RX detector formula to calculate the abnormal value of the spectral vector of the hyperspectral image after enhancing the abnormal target; 6) according to Outliers get detected results. The invention can obtain richer potential information in the hyperspectral image, increase the gap between the abnormal target and the complex background in the image, has the advantages of simple calculation and high detection accuracy, and can be used to detect the abnormal target in the hyperspectral image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a hyperspectral anomaly detection method, which can be used to detect abnormal targets in hyperspectral images. Background technique [0002] Hyperspectral images have rich spectral information and high spectral resolution, and have certain research value in the fields of target detection, classification, and recognition. Hyperspectral image anomaly detection is an unsupervised target detection method. When the prior information of the target and the background is unknown, the method judges whether it belongs to the abnormal point or the background target by comparing the difference between the detected point and the selected background spectral curve. Usually, prior knowledge of objects and backgrounds is difficult to obtain, so hyperspectral anomaly detection techniques have greater significance in practical applications. [0003] The classic anomaly detection a...

AI Technical Summary

Problems solved by technology

Although this method can reduce the computational complexity, it only considers the spectral features of the hyperspectral image and ignores the spatial features, so the detection accuracy is not high

Images