摘要
在实时三维目标跟踪系统中,KL距离自适应粒子滤波算法中距离阈值、小区域阈值以及其他参数的选取往往根据经验设置,如果参数设置不合适会降低跟踪精度和实时性。为此,设计一种3D点云目标跟踪系统。分析距离阈值和小区域阈值等参数对跟踪性能的影响,并给出自适应粒子滤波中参数与跟踪目标模型的关系。实验结果表明,与PCL_Tracking算法相比,该系统提高了三维目标跟踪系统的准确性和实时性。
In the real time 3D target tracking system,the distance threshold, small area threshold and other parameters in Kullback Leibler Distance( KLD) adaptive particle filtering algorithm can only be set according to experience,while improper parameter setting may reduce tracking accuracy and real-time performance. To solve this problem,a 3D target tracking system is designed. The influence of the parameters including distance threshold and small area threshold on the tracking performance is analyzed experimentally,and the relationship between the parameters and the tracking target model is given. Experimental results show that compared with PCL_ Tracking algorithm,the accuracy and real-time performance of the 3D target tracking system are better.
出处
《计算机工程》
CAS
CSCD
北大核心
2017年第9期304-309,共6页
Computer Engineering
基金
国家自然科学基金(61175094
61673304)
关键词
点云库
目标跟踪
粒子滤波
距离阈值
小区域阈值
Point Cloud Library(PCL)
target tracking
particle filtering
distance threshold
small area threshold