This paper studies regional tracking in the video sequences. A new method for tracking based on similar measure about Chebyshev distance is presented. The similar measure between the reference object and candidate object can be distinguished in the tracking. Then, the method of hypothesis testing is used for test, and the method of statistical test is used to judge right tracking or not. Finally, the result of the experiment shows that the method can avoid the probability distribution of the estimated density function, and comparative experiments are carried out to Kalman filtering in moving-known and to Camshift in moving- unknown respectively. The results show that the presented algorithm gets better tracking results.