Adaptive backgroud target tracking system based on TMS320DM642

This paper proposed an system of target tracking based on TMS320DM642 and discussed the overall hard design for the system. Considering the issue of poor performance for target tracking in complicated background especially when the background color is the similar with the target color. This paper combinded the method of weighted background histogram with the adaptive background Camshift algorithm, which effectively improve the tracing effects in the complex background. In the tracking process, the greedy algorithm is used for motion prediction, which reduce the numble of iterations and improve the stability of tracking. Using the Cache optimization based on EDMA, improve the efficiency of the algorithm and enhence the system´s real-time. The experimental results show this tracking system has better robustness and real-time property.