基于核相关滤波和卡尔曼滤波的目标跟踪算法研究  

Research of Object Tracking Algorithm Based on KernelCorrelation Filtering and Kalman Filtering

作　　者：甘志英 GAN Zhi-ying(Intelligence and Information Engineering College,Tangshan University,Tangshan 063000,China)

机构地区：[1]唐山学院智能与信息工程学院,河北唐山063000

出　　处：《唐山师范学院学报》2024年第6期60-65,共6页Journal of Tangshan Normal University

基　　金：河北省高等学校科学技术研究项目(QN2022186)。

摘　　要：针对遮挡情况目标跟踪经常产生漂移的问题,提出一种基于核相关滤波和卡尔曼滤波的目标跟踪算法。算法使用核相关滤波响应的峰值旁瓣比,判断遮挡。无遮挡时,以核相关滤波器为主跟踪器,获取目标位置,并修正卡尔曼滤波器;有遮挡时,将卡尔曼滤波器设置为主跟踪器,估计目标位置,保持核相关滤波模型。实验从跟踪速度、精度、成功率等角度与其它算法比较,结果表明该算法在遮挡情况下,有效改善跟踪效果,具有很强的鲁棒性。In view of the problem that target tracking often generates drift in the case of occlusion,an object tracking algorithm based on kernel correlation filter and Kalman filter is proposed.The algorithm u-ses the peak sidelobe ratio of the kernel correlation filter response to judge the occlusion.When there is no occlusion,the kernel correlation filter is used as the main tracker,the object position is obtained,and the Kalman filter is modified.When there is occlusion,the Kalman filter is set as the main tracker and the ob-ject position is estimated and keep the kernel correlation filter model not updated.The experiment com-pares the tracking speed,accuracy and success rate with other similar algorithms in qualitative and quanti-tative analysis,and the results show that the algorithm can effectively improve the tracking effect under the occlusion condition,and has strong robustness.

关 键 词：目标跟踪 核相关滤波器 卡尔曼滤波器 遮挡 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]