To address the challenges of drone detection under background clutter or occlusion, a correlation filter algorithm based on PSR discrimination was designed and implemented to achieve fast and robust drone tracking. In this paper, the KCF algorithm is used to detect drone in real time. The PSR detection confidence is used to estimate whether the drone is occluded or encountering a cluttered background, and the KCF algorithm parameters are adjusted accordingly to achieve more robust tracking. Experimental results show that the proposed algorithm effectively reduces the impact of background clutter and occlusion, can run in real time on a computer, and has high practical value for engineering applications.