Infrared Small Target Detection via Structure-Guided Diffusion Filter

Efficient detection of infrared small targets in complex backgrounds is of great significance for infrared search and track systems. Existing methods generally perform well in simple backgrounds but often suffer from false alarms in complex scenes due to inaccurate evaluation of the local background. To address this issue, this article proposes a target detection method based on structure-guided diffusion filter (SGDF). By modifying the diffusion coefficient and judgment condition of termination in the anisotropic diffusion filter (ADF), the model enables accurate detection of infrared small targets. In our diffusion scheme, the local region is vectorized. The description of the shape obtained from the vector space can effectively distinguish isotropic structures; moreover, the local background is evaluated by the composite vector. Furthermore, the diffusion process, guided by weights and aimed at the local mean, ensures both stability and efficiency. Extensive experiments conducted across various complex scenarios demonstrate that, compared with baseline methods, the proposed approach exhibits superior robustness and detection capability in challenging environments.