Infrared Small Target Detection Based on Directional Mean Difference and Compactness

The effective detection of small targets is a crucial task in infrared search and tracking applications. In this paper, we propose a method based on directional mean difference and compactness that will address some of the issues of the existing methods. First, the directional mean difference (DMD) is computed to extract the local bright regions. Then, the compactness of those regions is obtained and the DMD and compactness maps are integrated. Furthermore, the multi scale technique is applied for size adaptation and the small targets is segmented through an adaptive threshold. The simulation results confirm that the proposed method has achieved superior detection performance with a low computational load.