A tensor-based adaptive subspace detector for hyperspectral anomaly detection. Issue 8 (18th April 2018)

• Record Type: Journal Article
• Authors: Zhang, Lili; Cheng, Baozhi; Deng, Yuwei
• Is Part Of: International journal of remote sensing. Volume 39:Issue 8(2018)
• Page Start: 2366
• Page End: 2382
• Publication Date: 2018-04-18
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2017.1421795
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: In recent years, some methods based on tensor description have been proposed and perform well for hyperspectral anomaly detection (AD). However, these methods are mainly focused on the noise reduction or dimensionality reduction. In this article, a tensor-based adaptive subspace detection (TBASD) algorithm that can fully explore spatial–spectral potential without breaking the spatial–spectral structure is proposed for hyperspectral AD. First, the dual windows that are the guard window and the local neighbouring window, respectively, are employed. The centre pixel of the guard window is the test pixel, and the tensor block of the guard window is the test tensor block. The tensor blocks with the same size as the test tensor block are selected between the inner window and the local neighbouring window, and then in the whole hyperspectral imagery, respectively. Finally, the detection result is obtained using the tensor-based adaptive subspace detector. The experimental results demonstrate that the proposed TBASD can achieve a better performance when compared with the comparison algorithms.