摘要
联合概率数据关联算法具有良好的多目标跟踪性能,但其计算量会随着跟踪目标数和有效量测数的增多而呈指数增长,因此实时性差,难以在工程中应用。在保证准确率和精度的前提下减小确认矩阵的维数,提出了一种关联区域预处理的方法。对目标空间进行网格划分,通过网格的选取形成连通域,再对每个连通域中的目标采用联合概率数据关联算法,从而大量减少关联时间。仿真实验表明,基于网格连通的联合概率数据关联算法具有较强的实时性。
The Joint Probabilistic Data Association Algorithm has good capability in multi-target tracking, but the calculating amount would increase exponentially with the increasing of the number of targets and validated measurements, thus the algorithm has unsatisfactory real-time performance and can＇t be easily implemented in engineering.On the premise of guaranteeing the accuracy rate and precision, a new method on the preprocessing of the associating area is proposed to reduce the confirmed matrix dimension.The method partitioned the area according to the appropriate size, and chooses grids to form the connected area, then applies joint probability data association to different area individually, thus can decrease the associated time greatly.Simulations verify that the joint probabilistic data association algorithm based on grid and connection has better real-time performance.
出处
《电光与控制》
北大核心
2013年第9期34-36,92,共4页
Electronics Optics & Control
基金
海洋环境立体监测系统技术研究
关键词
联合概率数据关联
网格划分
连通域
关联时间
joint probabilistic data association
grid partitioning
connected area
associated time
