Tracking 2-D rigid targets with invariant constraints

An algorithm for tracking 2-D rigid non-point targets is presented. By introducing an interval turn model for center motion and considering a constraint of rigidity, the estimation of translation and rotation of the target can be obtained simultaneously when only measurements of target vertices with the presence of false alarms are available. A set of motion invariant coefficients are calculated along with the location estimates. Modeling vertices as a group of coupled members together with the coupled joint probabilistic data association (JPDA) algorithm improves tracking performance. This approach provides some insights to joint target tracking and identification (ID), and feature-aided tracking (FAT). (C) 2001 Elsevier Science Inc. All rights reserved.