Abstract: More and more applications nowadays need to use cameras to track moving objects automatically and more and more systems and methods have been developed to do motion track.However，one common fault of these systems and methods is that they fail to recognize moving objects when these objects are far away and environment affects their performance greatly.To address these issues，this article combines time differential and spatial gradation to develop an innovative motion track algorithm.The algorithm is not only capable of real-time tracking of human targets and automatic focusing，but it can also adapt to environment fairly well.

Key words: moving object tracking, time differential, spatial gradation

摘要： 利用摄像头对运动人像目标进行自动跟踪在越来越多的领域成为客观需要，国内外目前也已经有不少对运动目标自动跟踪的方法和系统。但是，这些方法和系统在运动物体距离较远时不能较好辨识目标，外界环境对于系统的性能也存在较大的影响。针对以上不足，将时域差分和空域分级结合，提出了一种实时的运动人像跟踪算法，使其不仅能够实现实时地跟踪运动人像目标，自动进行变焦，而且对外界环境有较好的适应性。

关键词: 运动目标跟踪, 时域差分, 空域分级