Generalized Data Association for Multitarget Tracking in Clutter

The objective of this chapter is to present an approach for target tracking in cluttered environment, which incorporates the advanced concept of generalized data (kinematics and attribute) association (GDA) to improve track maintenance performance in complicated situations (closely spaced and/or crossing targets), when kinematics data are insufficient for correct decision making. It uses Global Nearest Neighbour-like approach and Munkres algorithm to resolve the generalized association matrix. The main peculiarity consists in applying the principles of DezertSmarandache theory (DSmT) of plausible and paradoxical reasoning to model and process the utilized attribute data. The new general Dezert-Smarandache hybrid rule of combination is used to deal with particular integrity constraints associated with some elements of the free distributive lattice. The aim of the performed study is to provide coherent decision making process related to generalized data association and to improve the overall tracking performance. A comparison with the corresponding results, obtained via Dempster-Shafer theory is made. This work has been partially supported by MONT grants I-1205/02, I-1202/02 and by Center of Excellence BIS21 grant ICA1-2000-70016