Anomaly Detection Based on High-order Statistics in Hyperspectral Imagery

Abstract

According to the property of hyperspectral remote sensing data, a new anomaly detection algorithm based on high-order statistics is presented. The proposed algorithm used the augmented Lagrange multiplier (ALM) method to search for a projection that maximized the high-order statistics. They include normalized third central moment referred to as skewness and the normalized fourth central moment referred to as kurtosis, which measure the asymmetry and the flatness of the sample distribution respectively. They both are susceptible to outliers, so using these high-order statistics to detect anomalies may be effective. Comparison was made with a well-known anomaly detector, and results show that the proposed algorithm can effectively and reliably detect the small targets from hyperspectral images