Aiming at the problems of boundary effect and filter degradation in the correlation filter tracking algorithm in the high-speed scene of Unmanned Aerial Vehicle (UAV), this stady proposes an adaptive spatial-temporal regularization correlation filter algorithm for UAV tracking based on camera motion compensation. Firstly, an adaptive spatial regularization method is designed in the correlation filter to alleviate the boundary effect. Secondly, in order to prevent the degradation of the filter, a time regularization method and an optimization update strategy based on high-confidence samples are designed. At the same time, the ADMM optimization objective function is used to ensure the efficiency of the algorithm. Then, a fast scale filter is designed to achieve fast estimation of the target scale. Finally, a robust tracking uncertainty rule and a re-detector based on camera motion compensation are proposed, so that the target can be recovered in time after loss. The experimental results show that the precision and success rate of the proposed algorithm on the UAV dataset can reach 0.739 and 0.634, respectively, which are 1.65% and 6.9% higher than those of the ECO_HC algorithm. Compared with similar mainstream algorithms, it effectively improves the precision and success rate of tracking, and also shows good tracking performance when experiencing fast motion, target occlusion, etc.