基于GM(1,1)尺度估计的核相关性滤波跟踪方法

A Kernel Correlation Filtering Tracking Method Based on GM(1,1) Scale Estimation

针对在目标视觉跟踪过程中目标尺度变化导致跟踪精度下降的问题,提出了基于GM(1,1)尺度估计的核相关性滤波跟踪方法。在核相关性滤波跟踪算法的基础上引入GM(1,1)灰色预测模型对目标尺度进行预测,并设计尺度估计流程估计当前目标尺度,有效地解决了尺度变化明显导致跟踪性能下降的问题。最后利用OTB-100视频序列数据集测试算法的性能,测试结果表明在目标尺度明显变化时,该算法仍具有较好的跟踪性能,可实现稳定准确的目标跟踪,有效地提高了核相关性滤波跟踪算法的跟踪性能。

Aiming at the problem that the tracking accuracy decreases due to the change of target scale in the process of target visual tracking a kernel correlation filtering tracking method based on GM(11)scale estimation is proposed.The GM(11)grey prediction model is introduced to predict the target scale based on the kernel correlation filtering tracking algorithm and the scale estimation process is designed to estimate the current target scale which effectively solves the problem of tracking performance declining caused by obvious scale changes.Finally the OTB-100 video sequence dataset is used to test the performance of the algorithm.The test results show that the algorithm still has good tracking performance when the target scale changes significantly.It can realize stable and accurate target tracking and effectively improve the tracking performance of the kernel correlation filter tracking algorithm.