DIM Moving Target Detection using Spatio-Temporal Anomaly Detection for Hyperspectral Image Sequences

Dim moving target detection from hyperspectral image sequences, which contains temporal information as well as spectral information, has attracted researchers' interest for its crucial role in civil and military application. In this paper, we propose a novel spatio-temporal anomaly approach to solve the dim moving target detection problem. This approach calculates spatial anomaly map, temporal anomaly map using anomaly detection algorithm from spatial domain and temporal domain, respectively. To achieve motion consistency characteristic, this approach manages to generate the trajectory prediction map. After fusing the spatial anomaly map, the temporal anomaly map and the trajectory prediction map, target of interest can be easily detected from background. The proposed approach is applied to a test dataset of airborne target in the cloud clutter background. Experimental results confirm that the proposed approach can achieve a low false alarm rate as well as a high probability of detection.