嵌入式视觉运动目标跟踪技术研究

传统的PPM（Posterior Probability Measure）视觉运动目标跟踪算法,在目标发生尺寸变化时,会因跟踪窗不能自适应地改变尺寸而导致目标跟偏甚至跟丢的现象.因此本文在深入研究PPM跟踪算法的基础上,基于PPM跟踪算法的运算特点,提出了一种能袁征被跟踪目标尺寸大小的PPM缩放指标,进而设计了跟踪窗口自适应的PPM跟踪算法,实验显示,改进后的跟踪算法能得到较好的跟踪效果.基于TMS320DM642对嵌入式视觉目标跟踪系统的结构和功能进行设计,对改进后的PPM跟踪算法进行了DSP移植,最终实现了一套嵌入式视觉目标跟踪系统.

Study on visual target tracking embedded system

The traditional PPM （Posterior Probability Measure）algorithm is studied,based on the features of PPM,A new algorithm to estimate the visual target＇s scale is proposed.The algorithm overcomes the problem that the traditional PPM tracking algorithm often fails because of the rigid tracking window when target＇s scale changes.Experimental results show that the improved PPM tracking algorithm can be implemented in real-time and can adapt to changes of scale of the target.Then,based on TMS320DM642,the structure and function of the visual target tracking embedded system are designed,the improved PPM tracking algorithm is transplanted to the DSP.Finally,a real-time visual target tracking embedded system is implemented.