Abstract:Infrared small target detection technology holds significant application value in urban security, remote sensing surveying, and military guidance, but the combined interference from natural background clutter and artificial decoy flares severely degrades its detection accuracy and robustness. To address this issue, this paper presents an anti-decoy infrared small target detection algorithm integrating local contrast and directional derivative features. The proposed method first extracts local contrast reflecting spatial distribution characteristics and second-order directional derivative capturing edge structure features, then adaptively fuses these features to suppress background clutter and decoy interference while extracting small targets. To handle decoy occlusion scenarios, a feature mutation discrimination mechanism is introduced by calculating inter-frame feature differences to identify occlusion events, during which the Kalman filter predicts target positions to maintain detection continuity. After decoy separation, rapid target recapture is achieved. Comparative experiments on real-world infrared sequences with decoy interference demonstrate that the algorithm effectively suppresses complex backgrounds and decoys while achieving high detection accuracy and robustness.