Skip to Main Content
A modified deterministic annealing expectation maximization (MDAEM) algorithm has been proposed [D. Carevic, ldquoAutomatic estimation of multiple target positions and velocities using passive TDOA measurements of transients,rdquo IEEE Transactions on Signal Processing, vol. 55, no. 2, pp. 424-436, 2007] for the estimation of the motion parameters of multiple targets moving in cluttered three-dimensional (3D) underwater environments. In many situations, the number of targets in the measurements batch is not known, and the MDAEM algorithm is run for an inexact (guessed) number of the motion models, where, usually, the guessed number of models is larger than the true number of targets. This correspondence describes a detection procedure that determines which of the motion models estimated by the algorithm are related to the true targets in the measurements batch.