计算机工程与应用2017，Vol.53Issue(6)：141-144,173,5.DOI:10.3778/j.issn.1002-8331.1508-0022
基于Mean Shift的多模板目标跟踪算法
Multiple template moving object tracking based on Mean Shift
摘要
Abstract
Object tracking is one of the hot topics in computer vision. Although object tracking algorithms based on Mean Shift gain a lot of attention because of small calculation, real-time tracking and simple procedure, it can not deal well with the conditions of severe object mutation, frame loss or severe occlusion, which is needed to be improved. Different from the traditional moving object tracking method based on Mean Shift, this paper creates and maintains a diverse of template library providing more abundant description information to improve the tacking performance. Experimental results show that the new algorithm can track the moving object better in above situations, even keep a good robustness when complete object occlusion takes place.关键词
运动目标跟踪/MeanShift/多模板/模板库Key words
moving object tracking/Mean Shift/multiple template/template library分类
信息技术与安全科学引用本文复制引用
丁晓凤,尚振宏,刘辉,陈熙..基于Mean Shift的多模板目标跟踪算法[J].计算机工程与应用,2017,53(6):141-144,173,5.基金项目
国家自然科学基金(No.61462052,No.31300938,No.61262040). （No.61462052,No.31300938,No.61262040）
