摘要
提出了使用分类信息来改进传统的JPDA方法,在数据关联过程中综合使用了目标的运动学信息和目标的分类信息,提高了在跟踪波门内目标正确关联的概率,有效避免了航迹聚集现象.理论分析和仿真结果表明,尤其当多个临近目标的属性信息差别明显时,有目标分类信息辅助的JPDA方法可以显著提高跟踪波门内目标的正确关联概率,降低波门内其他目标/杂波的关联概率,使得被跟踪目标的信息更新以正确目标的信息为主.仿真结果验证了算法的正确性和有效性.
In this paper, a modified joint probabilistic data association with classification-aided avoiding track coalescence for multiple targets tracking is presented, target class information and kinematic information are integrated during the data association process to increase correct association probability. Theoretics analysis and simulation results show that classification-aided JPDA will increase the correct association probability of target in the predicted gate significantly, especially for multiple closely spaced targets with different attribute, avoid track coalescence and improve the track purity and tracking accuracy effectively. Simulation results show the algorithm validity.
出处
《武汉理工大学学报(交通科学与工程版)》
2009年第5期923-927,共5页
Journal of Wuhan University of Technology(Transportation Science & Engineering)
基金
"十一五"国防预研项目(批准号:513060302)
国家自然科学基金项目(批准号:60902071)资助
关键词
多目标跟踪
联合概率数据关联
分类信息
航迹聚集
multiple targets tracking
joint probabilistic data assoeiation
classification information
track coalescence