基于变尺度全局搜索的运动目标跟踪算法  

Moving object tracking algorithm based on variable scale global search

作　　者：张博[1] 张树群[1] 雷兆宜[1] 

机构地区：[1]暨南大学信息科学技术学院,广州510632

出　　处：《激光杂志》2015年第2期30-34,共5页Laser Journal

基　　金：广东省自然科学基金(32208005)

摘　　要：为了获得更加理想的运动目标跟踪效果,提出一种基于变尺度全局搜索的运动目标跟踪算法。首先对于没有发生严重遮挡运动目标,采用MS算法实现目标跟踪,然后当目标受到严重遮挡时,通过全局搜索策略缩小搜索窗尺度,逐渐逼近跟踪目标,实现对目标准确跟踪,最后采用多个实验对算法的性能仿真测试。仿真结果表明,相对其它的目标跟踪算法,本文算法不仅提高了运动目标的跟踪精度,而且加快运动目标跟踪速度,尤其对于严重遮挡目标跟踪问题,具有十分明显的优势,具有较强的鲁棒性。In order to obtain more ideal effect of moving target tracking,a novel object tracking algorithm based on variable scale global search is proposed in this paper. Firstly,First of all,for no serious occluded moving objects,MS algorithm is used to track for target,and then when the target is severely occluded,global search strategy is used to reduce the search window scale,gradually approaching the tracking target to achieve the target accurate tracking,finally,several simulation experiments are used to test performance of algorithm. The simulation results show that,compared with to other target tracking algorithms,the proposed algorithm not only improves the moving target tracking accuracy,but also accelerate the tracking speed of moving target,especially for serious occlusion target tracking problem,and it has very obvious advantages,so it has strong robustness.

关 键 词：运动目标跟踪 颜色直方图 均值偏移算法 自适应搜索窗 

分 类 号：TN911.7[电子电信—通信与信息系统]