Improving the RX Anomaly Detection Algorithm for Hyperspectral Images using FFT

Anomaly Detection (AD) has recently become an important application of target detection in hyperspectral images. The Reed-Xialoi (RX) is the most widely used AD algorithm that suffers from “small sample size” problem. The best solution for this problem is to use Dimensionality Reduction (DR) techniques as a pre-processing step for RX detector. Using this method not only improves the detection performance of algorithm, but also significantly reduces its runtime. This paper presents a novel DR technique that uses the Fast Fourier Transform (FFT) to perform the band reduction for RX detector. We compared the proposed method, named FFT-RX, with several well-known detectors such as RX, RX-UTD, KernelRX, PCA-RX and DWT-RX. These algorithms applied to two hyperspectral datasets acquired by both the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and Hyperspectral Mapper (HyMap) sensors. The evaluation of algorithms was based on Receiver Operation Characteristic (ROC) curve, visual investigation, and runtime of algorithms as well. Experimental results show that the proposed method improves the detection performance and runtime of RX detector significantly and has the best runtime and detection performance among all methods.