Characterization of anomaly detection in hyperspectral imagery

Purpose - The paper aims to characterize anomaly detection in hyperspectral imagery. Design/methodology/approach - This paper develops an adaptive causal anomaly detector (ACAD) to investigate several issues encountered in hyperspectral image analysis which have not been addressed in the past. It also designs extensive synthetic image-based computer simulations and real image experiments to substantiate the work proposed in this paper. Findings - This paper developed an ACAD and custom-designed computer simulations and real image experiments to successfully address several issues in characterizing anomalies for detection, which are - first, how large size for a target to be considered as an anomaly? Second, how an anomaly responds to its proximity? Third, how sensitive for an anomaly to noise? Finally, how different anomalies to be detected? Additionally, it also demonstrated that the proposed ACAD can be implemented in real time processing and implementation. Originality/value - This paper is the first work on investigation of several issues related to anomaly detection in hyperspectral imagery via extensive synthetic image-based computer simulations and real image experiments. In addition, it also develops a new developed an ACAD to address these issues and substantiate its performance.