Nonlocal Affinity-Based Robust Interference-Resistant Model for Infrared Small Target Detection

Abstract

Infrared small target detection (ISTD) is a fundamental component of infrared search and tracking (IRST) systems. The low-rank sparse decomposition (LRSD) method has become the mainstream of ISTD due to its broad applicability across various scenarios. However, certain sparse interferences in complex backgrounds may limit the effectiveness of these methods. To solve this problem, we propose a nonlocal affinity-based robust interference-resistant model (NARIRM) for ISTD. The model leverages the concept of affinity, which denotes the relationship between pixel regions, assuming that interference has stronger affinity with its neighbors than the target. The affinity values are achieved by reformulating an infrared image as the linear combination of foreground and background and using sparse decomposition results as constraints. A suppressor is then derived to reduce the impact of sparse interference by the affinity values. Experimental evaluation on public datasets demonstrates the proposed method outperforms several state-of-the-art techniques. The code is available at https://github.com/djk1997-jk/NARIRM