A Real-Time Algorithm of Data Association for Multitarget Tracking
Yangming Guo,Weihua Qin,Zhengjun Zhai,Hongmei Jiang
DOI: https://doi.org/10.3969/j.issn.1000-2758.2007.05.018
2007-01-01
Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University
Abstract:Under most circumstances, it is very difficult for Joint Probabilistic Data Association (JPDA) algorithm to achieve real-time multitarget tracking. We present an algorithm that we believe can do so in most circumstances. In the full paper, we explain our algorithm in some detail. In this abstract, we just add some pertinent remarks to listing the three topics of explanation. The first topic is: the construction of confirmation matrix. In this topic, we propose an approximation method for constructing the confirmation matrix through removing the small probability events using the right threshold values. To obtain the confirmation matrix, we arrived at an approximation probability matrix through modifying the association probability. The second topic is: the division of independent tracking space. In this topic, we present the mathematical division of the confirmation matrix according to the intersection area of the tracking gate of the targets to be tracked. The third topic is: the computation of association probability. In this topic, we use eqs. (6) through (10) in the full paper to compute the association probability of targets through attenuating the public measurement probability density in the same intersection area. Finally we conduct the Monte Carlo simulation of our improved JPDA algorithm and that of the JPDA algorithm. The simulation results, summarized in Tables 1 through 3 in the full paper, show preliminarily that: (1) compared with JPDA algorithm, our improved JPDA algorithm saves the CPU computing time greatly, namely superior in real-time to JPDA algorithm; (2) the tracking accuracy of our algorithm approximates that of JPDA when targets are moving in various modes of motion.