Automated multi-target tracking with kinematic and non-kinematic information

The authors address an automated multi-target tracking (MTT) problem. In particular, our study is focused on robust data association considering an additional feature and the reliable track management by avoiding track duplications. As the additional feature, the amplitude information is combined with position measurements to improve the performance of the data association so as to effectively distinguish target-originated measurements from clutters. Because of its form of signal-to-noise ratio (SNR), which is often fluctuated according to targets' aspect and effective radar cross section, the usage of the amplitude information is not straightforward. To reduce the certain level of uncertainty of the SNR, the authors propose the SNR estimation algorithm. Moreover, the authors avoid the track duplication problem to achieve the reliability of track maintenance. Specifically, the authors solve the problem by exploiting well-known mean shift algorithm to merge duplications into appropriate clusters. Simulation results demonstrate the effectiveness and high estimation accuracy of the proposed MTT filter compared to existing methods.