基于颜色的粒子滤波算法的改进与全硬件实现

摘要

Nowadays, the research on the implementation of color-based particle filters is facing the problem of tracking accuracy and real-time processing speed. This paper presents a modified color-based particle filter, which basing on the traditional SR resample algorithm, scatters the left particles causing by hardware circuits around the target. Results show that proposed particle filters improve the accuracy and robustness of object tracking.Moreover, the architecture of its full hardware implementation is depicted in the paper. The experimental study on FPGA indicates that the proposed color-based particle filters perform robust tracking at 72 FPS and with 7387 LEs (Logic Element) hardware cost. In addition to that, the framework of scalable distributed particle filters and its resample scheme are presented to adapt more complex scenarios, carrying out multi-feature and multi-target tracking.%传统基于颜色的粒子滤波算法在硬件实现中存在着跟踪效果不理想、实时性差等问题.该文结合硬件电路需要对基于颜色的粒子滤波算法进行了改进,在传统SR重采样算法的基础上将剩余粒子撒向目标点附近,以提高其在硬件环境下跟踪的准确性与稳定性.文中给出了改进算法的全硬件实现的电路架构,并在FPGA上完成了目标跟踪系统的实现.实验表明提出改进算法与硬件实现方案对简单背景环境下运动目标有着良好的跟踪效果,系统实时处理能力可达72 FPS,而硬件消耗为7387个逻辑单元.为进一步适应复杂的应用环境,在此粒子滤波器基础上给出了可扩展的分布式粒子滤波系统的架构以及相应的重采样策略,良好的并行性与可扩展性使得系统能够完成复杂背景环境下多特征、多目标的跟踪任务.