摘要
针对传统Mean Shift跟踪算法在目标存在背景干扰或遇到遮挡时,目标跟踪不准确的问题,提出了一种基于特征匹配运动检测预估的Mean Shift跟踪方法。采用Harris算法提取跟踪目标特征点进行运动定位检测,通过Kalman滤波器估计每一帧中目标迭代的起始位置,由Mean Shift算法从预估位置开始迭代搜索,最终实现目标跟踪。实验证明:提出的算法能够在遮挡的情况下对目标进行精准的定位检测,有效改善了复杂条件下的跟踪效果,具有较好的鲁棒性。
Aiming at the problem that traditional Mean Shift tracking algorithm cannot track target accurately on such these cases that background interference involved or target is occluded, a new tracking algorithm based on features matching motion detection prediction is proposed. The method extracts tracking target feature points by the Harris algorithm to realize motion location detection. The original iteration position is estimated by the Kalman filter,iterative searching of moving target starts at the location to shrink the searching scope and target tracking finally realized. The results of experiment indicate that the algorithm can detect location of tracking target accurately,effectively improve tracking effect while interference existed in the cluttering background, and the improved algorithm owns strong robustness.
作者
郑涛
郝行猛
陈梅
ZHENG Tao;HAO Xing-meng;CHEN Mei(School of Electrical and Automation Engineering,Hefei University of Technology,Hefei 230009,Chin)
出处
《传感器与微系统》
CSCD
2018年第7期135-137,141,共4页
Transducer and Microsystem Technologies
基金
工信部2016智能制造综合标准化与新模式应用项目(JZ2016GQBK0983)
