A modified PMHT tracking algorithm

Since the computational cost of Probabilistic Multi-Hypothesis Tracker (PMHT) is linear with respective to the number of targets and observations, significant work has been done to improve its performance. But the original PMHT is a measurement oriented algorithm which is hospital to clutters. To solve the problem, the target oriented Probabilistic Multi-Hypothesis Tracker (TO/PMHT) which takes account of the distance between measures and targets is provided. The TO/PMHT calculates the true association probabilities, and removes the weakness of hospital to clutters. The TO/PMHT is proved to be effective and validated by comparing to measurement oriented PMHT and traditional data association algorithm, such as MHT.