摘要
针对动态背景下,一般跟踪算法存在着被动跟踪的滞后或偏移的问题,提出了一种结合Kalman滤波器的Mean-Shift跟踪算法。对运动矢量进行预处理,得到一个平稳更能反映运动信息的矢量场;利用Mean-Shift搜索算法精确地确定对象位置;此基础上,利用Kalman滤波器算法进行运动估计预测,来确定运动的轨迹。实验表明:与现有的方法相比,该方法可从复杂场景中更准确地对运动对象进行轨迹的跟踪。
Dynamic background,general tracking algorithm has problem of passive tracking lag or shift,propose an algorithm which combines Kalman filter and Mean-Shift tracking algorithm. Preprocess motion vector to get a smooth vector field which can reflect movement information very well; using Mean-Shift searching algorithm to determine object position precisely; motion estimation prediction is carried out by using Kalman filter algorithm on this basis,to determine movement track. Experiment shows that this method can track trajectory more accurately from complex scene compared with existing methods.
出处
《传感器与微系统》
CSCD
2016年第8期137-140,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61202312)
中央高校基本科研基金资助项目(JUSRP51510)
关键词
全局运动补偿
压缩域
轨迹跟踪
global motion compensation
compressed domain
trajectory tracking
