Association of spatial information about targets is convention-ally based on measures such as the Euclidean or the Mahalanobis distance. These approaches produce satisfactory results when tar-gets are more distant than the resolution of the employed sensing principle, but is limited if they lie closer. This paper describes an association method combined with classification enhancing performance. The method not only considers spatial distance, but also information about class membership during a post-processing step. Association of measurements that cannot be uniquely associated to only one estimate, but to multiple estimates, is achieved under the constraint of conflict minimization of the combination of mutual class memberships.

With Monte Carlo simulations the performance of this new method is compared with a Kalman filter. This evaluation is per-formed in a multi-target environment with unknown correspondence between measurements and targets. The evaluation can not be only based on the root mean square error (RMSE) of the position estimate, but requires a performance assessment of the underlying target number estimation and the association. Therefore, two new measures are introduced.

The new method outperforms the Kalman filter approach with respect to association performance and RMSE.