Effective small DIM target detection by local connectedness constraint

The main drawback of conventional filtering based methods for small dim target (SDT) detection is they could not guarantee sufficient suppression ability towards trivial high frequency component which belongs to background, such as strong corners and edges. To overcome this bottleneck, this paper proposes an effective SDT detection algorithm by using local connectedness constraint. Our method provides direct control for target size, ensure high accuracy and could be easily embedded into the classical sliding-window based framework. The effectiveness of the proposed method is validated using images with cluttered background.