基于均值漂移和卡尔曼滤波的跟踪算法研究-《计算机技术与发展》

文章信息/Info

Title:: Research on Target Tracking Algorithm Based on Mean Shift and Kalman Filter

作者:: 吴佳家¹；高珏²；李敏¹；许华虎[3]; [1]上海大学计算机工程与科学学院；[2]上海大学计算机中心；[3]上海大学上海上大海润信息系统有限公司

摘要:: 为了寻找一种可以实际运用到学校监控系统的目标跟踪算法，文中对基本Mean Shift算法进行描述，并阐述算法的实际意义。 Mean Shift虽然以其不需要参数、不需要穷尽搜索区域等特性可以较好地实现目标跟踪，但是同时其也有不足，让其在某些跟踪条件下达不到很好的效果。为了使Mean Shift目标跟踪算法满足实际应用需求，通过添加核函数和增加权重的方式对基础Mean Shift算法进行扩展，并在分析Mean Shift算法的不足之后，提出一种Mean Shift与Kalman滤波相结合的目标跟踪算法。通过学校的视频监控平台对提出算法进行验证，实验结果表明，该算法可以有效地对目标进行跟踪。

Abstract:: In order to find a practical target tracking algorithm using in school monitoring system,the original Mean Shift algorithm is de-scribed in detail,and the practical significance of the algorithm is described too. Mean Shift algorithm can get better target tracking with-out arguments and searching no exhaustive area or other features,but at the same time it has shortcomings,so that it cannot get good re-sults. In order to meet the actual application requirements,the Mean Shift algorithm is extended by kernel function and weight and the meaning on this expansion is also explained,and in the analysis of the lack of the Mean Shift algorithm,an algorithm combined Mean Shift and Kalman filter was proposed. Through the school's video surveillance,the algorithm was verified,and the results showed that the proposed algorithm can effectively track the target.