Individual-gating-by-sorting in MHT
Viet Due Nguyen, Tim Claussen · International Conference on Information Fusion · 2014
An efficient individual gating method is described and tested. Compared to other approaches it does not solely calculate distances between the predicted and the actual measurement but checks for overlapping frames centered around them. In contrast to covariance ellipses, these frames are created only by using the main diagonal entries of covariance matrices. The overlap-check is done by sorting algorithms and binary search. Only measurements which pass the overlap-check are further processed by ellipsoid gating. In doing so, not only the computation time itself but also the complexity can be reduced compared to classical ellipsoid or rectangular gating. This Individual-Gating-by-Sorting is integrated into a Multi-Hypothesis tracker and applied to a multistatic sonar dataset. Results show outstanding performances. The gain compared to classical ellipsoid gating in terms of computation even further increases as the amount of data increases.