Weighted Divergence of Directionally Salient Features for Infrared Small Target Detection Enhancement

Infrared small target detection is widely used in fields such as dark and weak object detection, which is an important means of early warning for space environment safety. Drawing on entropy theory and incorporating the directionality of target diffusion, this study introduces a method for Infrared Small Target Detection Enhancement named Weighted Divergence of Directionally Salient Features (DSWD). Initially, this method assesses the local saliency of an image using a multi-layer nested window approach, and applies multidirectional spatial filtering to highlight point targets that radiate outward from the center to eight directions within the infrared spectrum, thereby enhancing the target against a complex backdrop. To address the disparity between the anisotropic distribution of saliency features and the homogeneous background distribution, this study employs a scatter measure to quantify the probabilistic distribution between the target area and the background, thus identifying regions of interest. Nevertheless, the efficacy of target extraction can be compromised by obfuscating elements, such as prominently illuminated backgrounds. To mitigate this, the study leverages an enhanced absolute mean square deviation technique for further target refinement, ensuring precise separation and extraction from the intricate background. The process culminates with the application of an adaptive threshold segmentation method for final target extraction. The experimental results show that for the variable space environment, the method proposed in this paper can accurately detect the weak targets in the space environment and reduce the false alarm rate, which is expected to be applied in the future space environment safety warning system or for reference.