基于模糊粒子滤波的视频跟踪方法研究

智能视频监控系统的基础就是对监控范围的视频序列中目标进行有效实时的跟踪,针对当前跟踪方法在光照突变、部分遮挡情况下的跟踪缺陷,本文提出了一种利用模糊粒子滤波（Particle Filter,PF）进行视频目标跟踪的新算法.该算法结合模糊控制理论,引入量测的模糊性来克服特征信息的不确定性,并建立一个基于背景建模技术的自适应高斯混合模型和基于序贯蒙特卡洛的跟踪算法（sequential Monte-Carlo-based）.在新算法的指导下,针对静止摄像机在光照条件改变以及目标发生部分遮挡的情况进行了深入的研究.通过多种模糊量测技术的比较和应用,本文方法实现了多个目标精确有效跟踪.

Video Tracking Method Based on Fuzzy Particle Filter

Intelligent video monitoring system is the basis of the monitoring range of targets in video sequences effectively real-time trackin. In view of the current tracking methods in the light mutation and partial occlusion cases of defects, this paperproposes a fuzzy Particle Filter （Particle Filter, PF） carries on the new algorithm of video target tracking. This algorithm combines with fuzzy control theory, introduces the measurement of fuzziness to overcome the uncertainty of the feature information, and establishes a adaptive gauss（ ~n mixture model based on background modeling and tracking algorithm based on sequential Monte Carlo. Under the guidance of the new algorithm, in view of the change of static camera in light conditions and the partial shade ,the problems are carried on thoroughly. Through the comparison of several fuzzy measurement technology and application, the method realizes the effective tracking multiple targets accurately.