Automatic target tracking algorithm based on NMI feature matching with Position Prediction and its DSP realization

Algorithm based on NMI feature matching and Position Prediction for automatic target tracking are develop for realize an DSP based real-time target tracking system. The target is firstly detected with a improved frame-difference method, and then the target area is segmented based on threshold segmentation and region growth. The NMI feature is used for matching. Finally a combining forecast algorithm is used in the tracking system. This tracking system is implemented with a DSP controller and software is accomplished on DSP/BIOS operation system. It is proved that it can be meet the needs of real-time stably tracking and overcome the disturbance brought by attitude variances and short time occlusion.