Abstract: Aiming at the problem that the traditional Kernel Correlation Filter（KCF） tracking algorithm will take the background information as a target to keep tracking but can not relocation the target, when the target is missing due to illumination variation, severe occlusion and out of view. On the basis of KCF, this paper introduces the outlier detection method as the target loss early warning mechanism, and proposes the target loss re-detection mechanism. This method detects the peak value of the response of each frame, if the abnormal peak value is found, the target is lost or will be lost. Then, the early warning mechanism warns, the target template update is stopped, the target loss re-detection mechanism is started, and search the target in the full frame. The experimental results show that the precision of the improved algorithm is 0.751, and the success rate is 0.579, which is 5.77% and 12.43% higher than that of the traditional KCF tracking algorithm, respectively. This solves the problem that the KCF tracker can not recover the target to keep tracking after the target is lost, the performance of the tracking algorithm is improved and the long-term tracking is realized.

Key words: Kernel Correlation Filter（KCF） algorithm, outlier detection, target loss, re-detection

摘要： 针对传统核相关滤波器（KCF）跟踪算法受光照变化、严重遮挡和出视野等因素影响，出现目标丢失现象时，跟踪器会将背景信息作为目标继续进行跟踪而不能重新定位目标的问题，在KCF的基础上，引入异常值检测方法作为目标丢失预警机制，同时，提出了目标丢失重检测定位机制。方法对每帧的峰值进行检测，发现异常峰值，则判定目标丢失或即将丢失，预警机制发出警告，停止目标模板更新，启动目标丢失重检测定位机制，在全帧搜索定位目标。实验结果表明，改进的算法精确度为0.751，成功率为0.579，较之传统KCF跟踪算法分别提高了5.77%和12.43%。解决KCF跟踪器在目标丢失后不能重新找回目标继续跟踪的问题，提升了跟踪算法的性能，实现了长期跟踪。

关键词: 核相关滤波器（KCF）算法, 异常值检测, 目标丢失, 重检测