Abstract: This paper proposed an adaptive tracking algorithm in which the Mean-shift algorithm is taken to obtain target position. Among this framework, the target model is updated by analyzing the results of SIFT points matching using least sctuares model. This algorithm can fast track target with the advantages that the template updating and the finding of scale coefficients arc implemented at the same time. The experimental results show that the algorithm is very robust to the cases that the scales of target vary largely.

Key words: Target tracking, Mean-shift, SIFT point, Scales variant, Least squares mode