Hyperspectral anomaly detection method and device, electronic equipment and readable storage medium

Abstract

The invention discloses a hyperspectral anomaly detection method and device, electronic equipment and a readable storage medium. The method comprises the following steps: firstly, carrying out pixel block division on a to-be-processed hyperspectral image to obtain a to-be-identified data pixel block corresponding to each pixel; and then inputting each data pixel block to be identified into a pre-constructed 3D convolution self-encoding model to obtain spatial-spectral feature information of each data pixel block to be identified, wherein the 3D convolution self-encoding model is used for performing encoding operation on an input data pixel block to obtain a spatial-spectral data block, and performing decoding reconstruction processing on the spatial-spectral data block to obtain spatial-spectral feature information. According to the hyperspectral anomaly detection method, each piece of spatial-spectral feature information is subjected to anomaly probability calculation by using a preset spatial-spectral joint anomaly method to obtain an anomaly detection result of each pixel, and finally, whether each pixel is abnormal or not can be judged according to the anomaly detection result, so that the hyperspectral anomaly detection accuracy of the remote sensing image can be effectively improved.

Description

technical field [0001] The present application relates to the technical field of image processing, in particular to a hyperspectral anomaly detection method, device, electronic equipment and readable storage medium. Background technique [0002] With the development of hyperspectral imaging spectrometers, the spectral information of remote sensing images has rapidly developed from panchromatic and multispectral to hyperspectral, bringing remote sensing technology into a new stage. Hyperspectral images have high spectral resolution, and the obtained image pixels have spectral information of hundreds of bands. Based on the different spectral characteristics of various ground features, hyperspectral remote sensing technology is widely used in technical fields such as ground feature recognition and classification, quantitative inversion, and abnormal target detection. Among them, hyperspectral anomaly detection has become a research hotspot in recent years because it does not r...

AI Technical Summary

Problems solved by technology

Or the method based on the self-encoding network sends each pixel to the network separately for feature extraction. Although useful features can be extracted, it will lead to the loss of the spatial relationship of each pixel in the original image, and the above spatial-spectrum combination method combines the spatial information. Considered separately from the spectral information, which destroys the spatial-spectral relationship, which will lead to low accuracy of the final hyperspectral anomaly detection

Images