论文部分内容阅读
在多目标多量测环境中,量测-航迹分配的知识一般不适于跟踪算法。在本文中,对量测-航迹分配问题,采用了严格的概率方法。不象在传统多假设跟踪(MHT)算法那样,把量测分配给航迹;相反地,使用由期望最大化(EM)方法导出的最大似然(ML)算法估计每次量测属于每个航迹的概率。这些量测-航迹的概率估计对于调用随机多假设跟踪(PMHT)算法的多目标跟踪器是固有的。PMHT算法在计算上是切实可行的,因为它既不要求量测-航迹分配的计算,也不要求修剪。
In a multi-target, multi-measurement environment, the knowledge of measurement-track assignment is generally unsuitable for tracking algorithms. In this paper, a strict probability method is adopted for the measurement-track assignment problem. Instead of assigning measurements to tracks as in the traditional Multi-Hypothesis Tracking (MHT) algorithm, conversely using the Maximum Likelihood (ML) algorithm derived from the Expectation Maximization (EM) method estimates that each measurement belongs to each Probability of track. These measurements - the probability estimate of the trajectory is inherent to a multi-target tracker that invokes a random multiple hypothesis tracking (PMHT) algorithm. The PMHT algorithm is computationally viable because it does not require measurement - calculation of track assignments or pruning.