针对单目标跟踪滤波性能评价标准,分析研究均方根误差、误差适配百分比、和方根误差、带有理论边界的状态误差、带有理论边界的新息序列、归一化估计误差平方和具有理论边界的归一化信息平方.针对多目标跟踪滤波性能评价标准,分析研究圆丢失概率、Hausdorff距离、Was-serstein距离和最优子模型分配距离.并讨论跟踪滤波算法时间复杂度的度量方法.
For single-target tracking, some metrics were analyzed, including root mean square error, per- centage fit error, root sum square error, state error with theoretical hounds, innovation sequence with theoretical bounds, normalized estimation error square with theoretical bounds, and normalized innova- tion square with theoretical bounds. For multi-target tracking, some other different metrics were also studied,including circular position error probability, Hausdorff distance, Wasserstein distance, and opti- mal subpattern assignment. Moreover, evaluation methods for time complexity in tracking filters were discussed.