A dedicated pipeline processor for target tracking applications

The task of predicting target location from sample to sample is approached from the self-management order paradigm (SMOP). This algorithm is concerned with the dynamic estimation of the optimal order of the time-dependent polynomial that best fits to the last observations of the target motion vector. Hardware-software trade-off is discussed for this application, taking into account the algorithms involved in the predictive tracking system. A pipelined multiprocessor architecture which achieves the predictive target tracking in near-real-time is presented