基于自适应多模板匹配的目标跟踪

为了解决在模板匹配中精度和速度之间的矛盾，本文结合了DSP便于实现算法和FPGA高速并行运算的优点，提出了一种基于DSP和FPGA协同作用的快速目标跟踪系统，主要在DSP上完成跟踪算法流程控制，在FPGA上实现耗时模块MAD的计算工作，建立两方通信，使该跟踪系统正常工作。将传统的模板匹配算法扩展成为多级模板匹配，并在跟踪算法中加入了抗旋转模板。最后，在Blackfin561系列Visual DSP 和EP3C25系列FPGA组成的系统平台上实现了该算法，并进行了相关的性能测试。实验表明，该系统可以准确、实时地跟踪运动目标，并能够有效地抗目标旋转和尺度变化。

Object tracking based on adaptive multi-template matching

In order to solve the contradiction between the accuracy and speed on template matching, this paper has presented a new tracking system, which combines advantages of DSP on implementing algorithm and FPGA on Parallel compute method. This system has established bilateral communication between DSP and FPGA, which can ensures the properly working of system. This system controls tracking algorithm processes on DSP and calculates the Time-consuming MAD on FPGA. This paper also extended the traditional template matching algorithm to become a multi-level template matching and included anti-rotating template tracking algorithm. Finally, this algorithm has been realized on DSP and FPGA. The experimental results indicate that this system can keep good performance on tracking objects even when Rotation and scale changes happened on the target.