Hyperspectral anomaly detection method based on linear background removal. Issue 1 (January 2021)

• Record Type: Journal Article
• Authors: Du, Huiting; Fan, Yanguo; Xu, Mingming
• Is Part Of: Journal of physics. Volume 1732:Issue 1(2021)
• Publication Date: 2021-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1732/1/012044
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: In view of the fact that the existing hyperspectral image anomaly detection algorithms only pay attention to distinguish between target and background by using spectral differences and ignore the correlation between pixels, and that local RX algorithm is easy to detect global non-anomalies as anomalies when the background is relatively complex, a hyperspectral anomaly detection method based on linear background removal is proposed. In this method, the edge information of hyperspectral image is obtained first, then the line part of edge image is extracted by Hough transform, and finally the local anomaly detection is carried out based on the linear background removal. By comparing the ROC curve and AUC value of this algorithm with RX algorithm and local RX algorithm, it shows that the algorithm of this paper can effectively reduce the false alarm rate of detection and has a better detection effect.