Abstract: To address the critical issue of target tracking loss caused by the sharp decline in the signal-to-noise ratio (SNR) under battlefield smoke and flare interference in infrared imaging guidance systems, a robust anti-interference tracking method based on multi-feature fusion is proposed in this paper. First, a physical model of infrared image degradation is established. Accordingly, a cascaded enhancement algorithm combining an improved dark channel prior (DCP) and wavelet thresholding is proposed to effectively suppress smoke and noise interference. Subsequently, an improved kernelized correlation filter (MFF-KCF) algorithm is constructed by fusing multidimensional features, including histogram of oriented gradients (HOG), local binary patterns (LBP), and histograms of optical flow (HOF). Furthermore, an anti-occlusion re-detection mechanism based on the peak sidelobe ratio (PSR) is introduced to achieve robust long-term tracking. Experimental results on the LSOTB-Smoke dataset under heavy interference conditions demonstrate that the proposed algorithm achieves a tracking success rate of 0.63, representing an improvement of over 60% compared to the original KCF algorithm. Additionally, it achieves a real-time processing speed of 60 fps on a domestic embedded platform. The proposed method effectively enhances the tracking robustness of infrared targets in complex environments, validating the application potential and engineering value of domestic high-performance computing platforms in the field of complex optoelectronic countermeasures.