基于颜色和边缘特征的均值迁移目标跟踪算法  

Mean shift target tracking algorithm based on color and edge features

作　　者：郑玉凤[1] 马秀荣[1] 赵晓琳[1] 王新闯[1] 郭亲弟[1] 

机构地区：[1]天津理工大学计算机与通信工程学院,通信器件教育部工程研究中心,薄膜电子与通信器件天津市重点实验室,天津300384

出　　处：《光电子．激光》2011年第8期1231-1235,共5页Journal of Optoelectronics·Laser

基　　金：天津市科技创新专项资金资助项目(10FDZDGX00400)

摘　　要：综合颜色特征对旋转、变形不敏感和边缘特征对光照、颜色不敏感的优点,提出了一种基于颜色和边缘特征的均值迁移目标跟踪算法,提高了跟踪算法的鲁棒性。利用加权求和的方法获得联合特征的Bhattacharyya系数,以此作为衡量目标模型与候选目标相似程度的依据。实验结果表明,在光照变化时,本文算法的跟踪误差大约是基于颜色特征的均值迁移目标跟踪算法的1/3;在发生旋转、变形致使基于边缘特征的均值迁移目标跟踪算法丢失目标时,本文算法仍能很好地跟踪目标。A mean shift tracking algorithm based on color and edge features is proposed. The proposed algorithm combines the advantage of color feature that it is not sensitive to rotation and deformation and the advantage of edge feature that it is not sensitive to illumination changes and color changes. The robustness of the tracking algorithm is improved. As a measure of similarity between the target model and the candidate target, the Bhattacharyya coefficient of the combined features is obtained with weighted summation. The proposed algorithm is compared with mean shift algorithms based on color and based on edge feature respectively. Experimental results show thai：when illumination changes, the tracking error of the proposed algorithm is about 1/3 of that of the mean shift tracking algorithm based on color feature. When rotation or deformation occurs, the target may be lost with the mean shift tracking algorithm based on edge feature,while, the proposed algorithm can still work well.

关 键 词：目标跟踪 均值迁移 颜色特征 边缘特征 BHATTACHARYYA系数 

分 类 号：TN919.8[电子电信—通信与信息系统]