Comparative evaluation of hyperspectral anomaly detection methods in scenes with diverse complexity oecd

Anomaly detection in hyperspectral data has received a lot of attention for various applications and is especially important for defence and security. The aim of anomaly detection is to detect pixels in the hyperspectral data cube whose spectra differ significantly from the background spectra. Many types of anomaly detectors have been proposed in literature. They differ by the way the background spectra are defined and described and by the method used for determining the difference between the pixel under test and the estimated background characteristics. The most well-known anomaly detector is the RX detector. Several detectors have been derived from the basic RX detector. On the other hand methods based on image segmentation have also been introduced. These are particularly useful in areas characterised by a highly structured background (e.g. urban scenes). The current paper presents a comparison of the results obtained by representative examples of two classes of anomaly detector: the RX-family of detectors and the segmentation-based detectors.