基于颜色纹理特征的均值漂移目标跟踪算法  

An improved Mean-shift tracking algorithm based on color and texture feature

作　　者：戴渊明[1] 韦巍[1] 林亦宁[1] 

机构地区：[1]浙江大学电气工程学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2012年第2期212-217,共6页Journal of Zhejiang University：Engineering Science

基　　金：国家"863"高技术研究发展计划资助项目(2008AA042602);国家自然科学基金资助项目(60704030);中央高校基本科研业务费专项资金资助项目

摘　　要：针对经典均值漂移跟踪算法采用单一的颜色特征对目标进行跟踪检测存在的不足,提出一种将纹理特征与颜色特征相结合的改进均值漂移目标跟踪算法.该算法首次提出特征联合相似度的概念,通过均值漂移算法联合相似度的最大化计算,正确快速地获取新一帧图像跟踪目标的位置.实验结果表明,该算法具有更高的可靠性,同时满足一般目标跟踪任务的实时性要求.Aimed at the defect of classic Mean-shift tracking algorithm which is vulnerable to similar background inference for using single color feature,an improved color and texture features combined Mean-shift tracking algorithm is presented.Feature joint similarity was introduced for the first time.By maximizing the integrated similarity using Mean-shift algorithm the center of target in the new frame can be obtained accurately and rapidly.Experiments show that the proposed algorithm provides more reliable performance while satisfying the real-time requirements of general target tracking tasks.

关 键 词：目标跟踪 均值漂移 联合相似度 颜色特征 纹理特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]