An Adaptive Kernel Method for Anomaly Detection in Hyperspectral Imagery

An adaptive kernel method is proposed for anomaly detection in hyperspectral imagery. The conventional RX anomaly detector suffers from high false alarm rates and low probability of detection due to the assumptions that the local background is Gaussian and homogeneous. Many kernel base anomaly detector are of good nonlinear detection performance, but parameter estimation of the kernel function is difficult for these methods. In this paper, we proposed a adaptive parameter estimation method based on the sum of each spectral band standard deviation for the background clutter pixels and applied it in the kernel RX detector. The kernel function parameters estimation can be obtained along with the shifting of the background clutter pixels automatically so we can avoid a large number of experiments for the parameters determination. Numerical experiments are conducted on real hyperspectral imagery collected by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). Using receiver operating characteristic (ROC) curves, the results show the improved performance and reduction in the false-alarm rate.