基于N记忆簇标记SMC-PHD滤波的量子跟踪

使用PHD粒子滤波跟踪量子以满足特定生物信息学应用的需求,并结合簇标记技术实现多目标航迹关联.针对簇标记技术存在的"标签漂移"问题,提出一种N记忆簇标记方法用于控制算法对不稳定目标的敏感度.它通过记忆和对比过去N个时刻的目标状态可以有效降低目标频繁出没对目标估计和跟踪带来的干扰.实验结果表明该方法可以对具有高随机动态的量子目标实现较准确、鲁棒的航迹跟踪.

Quantum dots tracking with N-memory cluster-indexed SMC-PHD filter

A Cluster-Indexed SMC-PHD filter was presented to track quantum dots for certain biological purpose.It committed track association by labeling targets with track numbers.One problem of it is the labels of targets shift often when dealing with unsteady targets.The label shifting made it hard to obtain long-term tracks.A novel N-memory cluster-indexed SMC-PHD filter was proposed to control the sensitivity to targets birth and death.It checked the tracks back N time intervals to find out if any one of them could be associated with the currently emerging targets,and thereby,rejected false detection alarm resulting from the go-and-back of unsteady targets.Results show the method successfully track quantum dots of high dynamics.