基于均值偏移算法的目标跟踪系统及其在Blackfin上实现

摘要

描述了一个以Blackfin DSP为核心的图像实时采集和高速处理的运动目标跟踪系统设计.实践了一种快速匹配的均值偏移(Meanshift)跟踪算法,给出了在复杂背景下跟踪汽车的试验结果.实验证明该系统性能可靠,跟踪效果较好,并有较高的实时性.%In this paper, a real-time moving object tracking system of high-speed image sampling and processing with Blackfin DSP as the core component is presented. An algorithm of object-tracking based on Meanshift is implemented. At last, the experimental results under a complicate background are also shown. It is proved to be a reliable, effective and high real-time.