基于灰色关联分析的目标跟踪器性能评估方法

Performance evaluation method of target tracker based on grey correlation analysis

跟踪器在多因素多水平序列测试时,各跟踪性能指标的数据较多,难以直接获取评价算法跟踪性能的有效信息。为解决该问题,首先对跟踪性能指标进行筛选和改进,选择区域重叠度和中心位置误差两个指标反映跟踪效果,然后采用灰色关联分析的方法分析各个指标数据,通过分析跟踪结果,提取跟踪结果间的关联度信息,并以此信息计算跟踪器的跟踪精度和鲁棒性。实验表明,该方法计算的关联度信息能够准确反映算法性能变化趋势,从而可以较好地衡量和比较算法性能。

In the multi-factor and multi-level sequence testing,the tracking performance index has a lot of data,it is difficult to directly obtain the effective information to evaluate the tracking performance of the algorithm.In order to solve this problem,firstly,the tracking performance index is screened and improved,and two indexes,the region overlap degree and the center position error,are selected to reflect the tracking effect.Then the grey correlation analysis method is used to analyze the data of each index.By analyzing the tracking results,the correlation degree information between the tracking results is extracted,and the tracking accuracy and robustness of the tracker are calculated with the information.Experiments show that the correlation degree information calculated by this method can accurately reflect the performance trend of the algorithm,so that the performance of the algorithm can be well measured and compared.