Tracks extraction of the Probability Hypothesis Density filter for survival targets

The outcome of probability hypothesis density or cardinalized probability hypothesis density filter is a random finite set at each time step. However, the interesting thing in practice is the tracks of survived targets. In order to get the tracks of survived targets, data association technique should be used to pair existing tracks with the elements of the outcome random finite set at each time step. In this paper, a data association approach based on the Wasserstein distance for performance evaluation of multi-target tracking filter is proposed. The dasiatransportation matrixpsila of the Wasserstein distance is obtained through an optimal assignment algorithm, and the data association matrix can be obtained from the dasiatransportation matrixpsila. Simulation reveals that data association approach is successful.