摘要： 针对视频序列中缓变运动目标检测与跟踪的问题，改进加权累积差分方法，设计一种快速检测缓变运动目标的权值分配算法。在检测出的运动目标区域中提取特征点，利用最大距离原则确定特征点的匹配范围，根据视差一致性原则，提出非回溯的误匹配检测方法。实验结果表明，该方法能有效提高跟踪的效率和精度。

关键词: 累积差分图像, 运动目标检测, 跟踪, 特征点

Abstract: Aiming at the problem of tardo moving object detection and tracking, this paper proposes a new weighted accumulative difference method and works out a weights setting algorithm to fast detect the tardo moving object. It obtains the interest point from the moving area detected, fixes the matching range with the maximum velocity principle and works out a non-traceback mis-matching detection algorithm with the consistency principle. Experimental result shows that the method can enhance the efficiency and precision of our tracking system.

Key words: Accumulative Difference Picture(ADP), moving object detection, tracking, interest point

中图分类号: 

• TP37