In response to the large tracking error of anti-aircraft fire control systems when tracking complex maneuver of small and micro unmanned aerial vehicle (UAV), an interacting multiple model-improved strong tracking unscented Kalman filter (IMM-ISTUKF) algorithm based on adaptive transition probability matrix (ATPM) was proposed. The proposed algorithm combined a multiple fading factor to enhance the filtering performance of STUKF, and adaptively adjusted each element of the Markov transfer probability matrix using posterior probability and likelihood function information. A two-stage ATPM update mechanism based on model polarization correction and model transfer acceleration was designed to suppress interference from mismatched models during the stable phase, and accelerate model switching during the transition phase, thereby reducing state estimation error. Simulation results demonstrate that the proposed algorithm achieves higher tracking accuracy than traditional IMM and improved IMM algorithms, providing technical support for anti-UAV operations of anti-aircraft gun system.