Anomaly detection of hyperspectral imagery using differential morphological profile

Anomaly detection has been an interesting topic in hyperspectral imagery. Most anomaly detection methods use spectral information for detecting targets. In this paper we propose a method which uses both spectral and spatial information for detecting anomalies: differential morphological profile anomaly detection (DMPAD). This method utilizes principal component analysis (PCA) and differential morphological profile (DMP) to extract spectral and spatial information from hyperspectral image (HSI), respectively. The experimental results confirm DMPAD method's superiority to three mostly used anomaly detection methods, namely PCA, fisher linear discriminant (FLD) and Reed-Xiaoli (RX) methods. DMPAD detects more accurate targets and less false alarm in comparison with competing method.