Tracking of crossing targets with forward looking infrared imaging sensors

Abstract

An algorithm is presented for the tracking of crossing targets using the centroid measurement and the centroid offset measurement for the distributed image formed by the targets. The measurements are obtained by a forward-looking infrared imaging sensor. Overlap of target images occurs when the lines of sight of the targets cross. The resulting centroid measurement is a single merged measurement which is a linear combination of the centroids of the individual targets under consideration. Also the ensuing image correlation coefficient matrix for two frames is multimodal and care has to be taken to associate the derived offset measurements with a particular target properly. The overlap of the images for several sampling times causes a dependence of the state estimation errors for the two targets, which has to be taken into account. The joint probabilistic data association merged-measurement coupled filter is employed for state estimation. It performs filtering in a coupled manner for the targets with common measurements. Two filters are examined, one assuming that the displacement noise is white and the other one modeling it correctly as autocorrelated. The latter is shown to yield substantially better performance. The simulation results presented validate the performance predictions for the proposed algorithm.