Extended Target Probability Hypothesis Density Filter with Multiple Matching Criteria

The Extended target Probability Hypothesis Density (ET-PHD) filter is well-suited for tracking multiple targets with complex shapes or multiple components. However, the accurate calculation of the corrector of the filter requires consideration of all partitions of the measurement set, making the ET-PHD filter computationally impractical without approximations. A major challenge in implementing ET-PHD filters is maintaining computational efficiency and scalability for real-time tracking applications. To tackle this issue, researchers have developed various measurement set partitioning algorithms to simplify the filter's computation. However, most existing algorithms rely solely on the current measurements and only partition the measurement sets based on the distance between these measurements. In this study, we introduce multiple matching criteria to enhance the tracking accuracy of the ET-PHD filter and reduce computational complexity. By utilizing multiple matching of measurements to targets, we aim to effectively differentiate between measurements from different targets. Then, the partitioning algorithm and the ET-PHD filter with multiple matching criteria are proposed. The simulation results demonstrate that incorporating multiple matching criteria into the ET-PHD filter leads to improved tracking performance and reduced computational complexity.