Abstract: This paper develops a new general Viterbi MHT algorithm for multitarget tracking. A measurement “gating” is used in the algorithm, and the target associates the measurement that is in the “gating”. The method can decrease the number of hypothesis, reduce the computational burden of the algorithm, and benefit for pruning/merging. MAP path costs are computed by using Kalman filters and priori probabilities. The algorithm is sequential, and can deal with missed detections, false alarms and the number of track target. It can provide a list of best track sets.

Key words: multitarget tracking, pruning/merging, data association

摘要： 使用改进的Viterbi算法用于多目标跟踪，引入测量“门限”，使所跟踪的目标仅与“门限”内的测量值关联。该方法能够减少假设的个数、降低算法的计算负担，有利于对MHT算法进行剪枝和合并。用Kalman 滤波和先验概率计算各目标的最大后验概率。该算法是连续的，能够处理丢失的探测、虚警以及跟踪目标的数量，提供一系列最好的跟踪目标集。

关键词: 多目标跟踪, 剪枝/合并, 数据关联

CLC Number: 

• TN953