基于LK光流跟踪法的有效目标点增强跟踪  

Valid points reinforced tracking based on LK optical flow algorithm

作　　者：蒋菱[1] 程赓[1] 

机构地区：[1]四川大学计算机学院(软件学院),四川成都610065

出　　处：《微型机与应用》2015年第6期45-49,共5页Microcomputer & Its Applications

基　　金：国家自然科学基金(60903118;61103137);国家科技支撑计划(2012AA011801)

摘　　要：为了提高基于几何形态的物体跟踪方法的准确率,详细分析了基于几何形态的物体跟踪方法下目标点的正确性,并根据分析的结果 ,在对目标物体采样时选取更容易被跟踪到的目标点,对于跟踪的结果 ,排除由于背景影响产生的无效目标点,从增强有效点和排除无效点这两方面入手,提出了基于LK光流法的有效点增强跟踪算法。实验证明,提出的算法能够更加有效地跟踪物体,尤其是当目标快速运动或者发生形变时,能够有效地减少跟踪目标的中心错误率。To improve the accuracy of object tracking method based on the geometric shape, this paper analyzes the correctness of the target points in geometric-based tracking method in detail In accordance with the result of analysis, for the sampling of the target object, selecting the points which can be tracked more easily; for tracking results, excluding the invalid points coused by background. From the aspects of enhancing the valid points and excluding the invalid points, this paper proposes a valid points reinforced tracking method based on LK optical flow algorithm. Experiments show that proposed method performs better, and can effectively reduce the center error rate of the object especially when the target moves fast or deformation occurs.

关 键 词：目标跟踪 LK光流法 有效点增强 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]