The optimal 2D multiframe detector/tracker

We detail in this paper the implementation of the optimal Bayes multiframe detector/tracker for rigid objects moving randomly in two-dimensional (2D) finite grids. We present 2D models for target signature and target motion t hat, build an integrated framework for detection and tracking. We model the background clutter by 2D correlated noncausal Gauss-Markov fields of arbit rary order. Ey exploring the structure of the signature, motion, and clutte r models, we indicate how substantial computational savings can be achieved in the implementation of the algorithm. The detection performance of the p roposed Bayes scheme is evaluated through Monte Carlo simulations. The resu lts show significant performance gains of over 6 dB in peak signal-to-noise ratio when the optimal multiframe detector is compared to the optimal sing le frame likelihood ratio test (LRT) detector.