Non-linear anomaly detection for hyperspectral image

Abstract

Anomaly detection finds data samples whose signatures are spectrally distinct from their surrounding data samples. One area of interest regarding hyperspectral images is anomaly detection. Anomaly detection becomes increasingly important in hyperspectral image analysis. Hyperspectral imagers can now uncover many material substances which were previously unresolved by multispectral sensors. Many anomaly detectors have been developed and most of them are based on the most widely used Reed-Yu’s algorithm, called RX detector (RXD). Principal component analysis (PCA) allows linear dimensionality reduction and feature extraction for hyperspectral remote sensing data. Kernel PCA (KPCA) is the nonlinear form of PCA, which better exploits the complicated spatial structure of high-dimensional features. In this paper, we first use PCA-global RX (PCA-GRX) algorithm for anomaly detection. Kernel Principal Component Analysis (KPCA)-GRX algorithm is investigated for anomaly detection from hyperspectral remote sensing data. Experimental results presented in this paper confirm the usefulness of the KPCA-GRX for the analysis anomaly detection of hyperspectral data. The detection accuracy increases with the proposed approach.