基于粒子滤波与Mean shift结合算法对移动目标的跟踪分析

发布时间：2018-04-01 06:57

本文选题：粒子滤波　切入点：移动目标　出处：《新疆大学》2017年硕士论文

【摘要】：视频监控对目标展开跟踪的计算是近些年来在计算机领域中应用非常广泛的热点项目之一。该项技术不论是在军事领域上的应用还是在民用领域上的应用,都有着非常重要的研究价值。随着科技的进步,对粒子滤波的使用和研究开始得到了越来越多领域的重视。并且从近些年来看,粒子滤波算法已经被广泛地应用到了视频监控中,对要追踪的目标进行跟踪。在视频监控中对移动目标进行监控的过程是非常复杂的,其中不仅包括了目标物体的动态变化,还包括了移动目标之间可能发生的遮挡、合并以及分离等情况。因此,本文的研究目的与意义旨在即粒子滤波算法的技术应用之上,加强对视频中移动目标的出现进行应有的跟踪和必要的分析,解决好目标在移动过程中可能出现的复杂的情形,然后利用好粒子滤波的计算方法,从而实现对移动目标在视频中被成功跟踪及进一步追踪的观察目的。本文在行文的过程中对视频中移动的目标主要运用了粒子滤波算法与Mean shift算法,并最终对这两种算法进行了跟踪实验效果对比。在正文整体结构上:第一,本文先介绍了几种比较常见的目标跟踪算法,并对这些算法的特点进行了相关分析与说明;第二,本文论述了与粒子滤波相关的理论框架以及利用粒子滤波对移动目标进行跟踪的算法,并且用具体的视频跟踪实验说明了粒子滤波在跟踪视频中对移动目标进行跟踪的可靠性;第三,介绍Mean shift算法,并且通过对同一个视频图像中的移动目标进行跟踪实验,对比发现这两种算法的各自特点,为后续工作做准备;第四,在先前内容所研究的基础上,总结出了本文的核心内容“基于粒子滤波算法与Mean shift算法结合使用基础之上”的对视频中移动目标进行跟踪和计算的算法。这个算法可以有效的对被跟踪的目标在遮挡发生之后再次对被跟踪目标进行跟踪处理,能在一定程度上对移动目标在发生遮挡时容易造成跟踪丢失的现象起到弥补作用,同时该算法也能够在最快的时间内鉴别出移动目标可能发生的情况,从而进行结合算法的选择性应用,继而来应对移动目标被遮挡的现象,并且还可以有效地降低视频噪声及杂波的干扰,从而有效地提高视频跟踪系统对移动目标的跟踪。
[Abstract]:The computation of video surveillance and target tracking is one of the most popular projects in the computer field in recent years. This technology is used in both military and civilian fields. With the development of science and technology, the use and research of particle filter have been paid more and more attention. And in recent years, Particle filter algorithm has been widely used in video surveillance, tracking the target to be tracked. In video surveillance, the process of monitoring moving target is very complicated, which includes not only the dynamic changes of the target object, but also the dynamic change of the target object. It also includes the possible occlusion, merging and separation between moving targets. Therefore, the purpose and significance of this paper is to apply the particle filter algorithm. We should track and analyze the moving targets in video, solve the complex situations that may occur in the moving process, and then make good use of the particle filter calculation method. In this paper, particle filter algorithm and Mean shift algorithm are mainly used to track moving target in video. Finally, the experimental results of the two algorithms are compared. In the overall structure of the text: first, this paper introduces several common target tracking algorithms, and the characteristics of these algorithms are analyzed and explained. In this paper, the theoretical framework related to particle filter and the algorithm of moving target tracking by particle filter are discussed, and the reliability of particle filter in tracking moving target in tracking video is illustrated by specific video tracking experiment. Thirdly, the Mean shift algorithm is introduced, and the experiment of moving target tracking in the same video image is carried out, and the characteristics of the two algorithms are compared to prepare for the follow-up work. Fourth, on the basis of the previous research, This paper summarizes the algorithm of tracking and calculating moving target in video, which is based on particle filter algorithm and Mean shift algorithm. This algorithm can effectively track the target to be tracked. Track the tracked target again after the occlusion has occurred, To a certain extent, it can make up for the phenomenon that the moving target is likely to lose track when occlusion occurs. At the same time, the algorithm can also identify the possible situation of moving target in the fastest time. Therefore, the selective application of the combined algorithm can deal with the occlusion phenomenon of moving target, and it can also effectively reduce the noise and clutter interference of video, thus effectively improve the tracking of moving target in video tracking system.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】