一种基于Meanshift算法的目标跟踪改进  

An Improved Meanshift Algorithm for Target Tracking

作　　者：朱钾 李化[1] 

机构地区：[1]太原理工大学信息工程学院,山西晋中030600

出　　处：《中北大学学报（自然科学版）》2016年第6期633-637,665,共6页Journal of North University of China(Natural Science Edition)

基　　金：山西省自然科学基金资助项目(2014011019-1)

摘　　要：针对遮挡及相似干扰问题,在Meanshift算法的基础上提出了一种新的改进算法.该算法有效地将Meanshift算法与Kalman滤波算法结合,准确确定目标质心位置,并根据目标质心位置及前一帧搜索窗口边界自适应地调整当前搜索窗口,使其与目标位置更匹配,跟踪结果更加准确、可靠,同时也减小了环境等因素对算法的影响.实验对比结果表明,改进算法具有良好的稳定性和自适应性,可有效解决大面积遮挡和相似干扰问题.Aiming at the problem of occlusion and similar interference,a new improved algorithm based on the Meanshift algorithm was proposed.The algorithm effectively combined the Meanshift algorithm with the Kalman filtering algorithm to accurately determine the location of the target centroid.According to the target centroid position and the front frame search window boundary,the current search window boundary is adjusted adaptively to match the target position.In this way,the tracking results are more accurate and reliable,and the influence of the environment and other factors on the algorithm are also reduced.Through the experimental analysis,the improved algorithm which has good stability and adaptability can effectively solve the problem of large area occlusion and similar interference.

关 键 词：目标跟踪 MEANSHIFT算法 Kalman滤波算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]