Background Purification Framework With Extended Morphological Attribute Profile for Hyperspectral Anomaly Detection

Hyperspectral anomaly detection has attracted extensive interests for its wide use in military and civilian fields, three main categories of methods have been developed successively over past few decades, including statistical model-based, representation-based, deep-learning-based methods. Most these algorithms are essentially trying to construct proper background profiles, which describe the characteristics then identify pixels that do not conform profiles as anomalies. Apparently, crucial issue is how build an accurate profile; however, constructed by existing enough. In this article, a novel universal purification framework with extended morphological attribute proposed. It explores spatial characteristic image removes suspect from obtain purified background. Moreover, detectors covering different also developed. The experiments implemented on four real hyperspectral images demonstrate effective, universal, suitable. Furthermore, compared other popular algorithms, perform well terms accuracy efficiency.