Comprehensive Track Unswappinng for Improved Tracker Performance

In practical surveillance systems, multiple-target tracking can suffer from undesirable effects such as track breakages and track swaps. Track stitching or track segment association (TSA) algorithms have been proposed in the literature to stitch broken tracks deemed to have originated from the same target across time and to improve track continuity. Measurements from multiple neighboring targets may fall within the validation gates of one another causing association errors that may eventually lead to not just track breaks but also track swaps. Therefore, TSA alone is insufficient to improve the overall tracker performance, as it considers only the broken tracks but not the continuous ones that might have swaps among themselves or with other broken tracks. To mitigate the effects of track swaps, this article proposes an algorithm that detects and resolves possible track swaps using kinematic and non-kinematic—classification and amplitude—information. Track swap detection involves identifying the most likely instant of track swap occurrence. Further, the proposed algorithm is extended to stitch broken track segments (as in the standard TSA) and those tracks that are algorithmically broken due to the detection of possible swaps. Simulation results demonstrate the effectiveness of the proposed algorithm in resolving track swaps and thereby improving track purity and the overall tracker performance.