林场复杂背景中相关跟踪算法的分析及实现-analysis and implementation of correlation tracking algorithm in complex background of forest farm.docxVIP

摘 要随着科学技术的发展，相关跟踪技术的研究在森林安防、安防监控、机器人 等技术发展中有着重要的应用前景和理论意义。在实际目标跟踪过程中，特别是 复杂背景中的目标跟踪，由于目标机动性高，对目标的跟踪相对困难。在此跟踪 条件下，相关跟踪算法是最常用的一种，它易于实施、跟踪精度高，但其计算量 大、易累积误差，使其不利于在实际跟踪系统中使用，针对相关跟踪算法存在的 问题，本文对其进行了分析和研究。本文主要研究基于图像灰度的相关跟踪算法，并在此基础上构建了一个实时 目标跟踪系统的实验验证平台，实现对复杂背景中运动的目标进行实时跟踪。首 先，在分析、比较现有的相关跟踪算法的基础上，采用跟踪性能较好的去均值相 关跟踪算法；其次，通过去均值相关跟踪算法公式的变形和跟踪波门的设置相结 合，减少相关度量的总体计算量；再次，将本文提出的模板更新策略、目标失锁 判断及重新捕获准则与相关跟踪算法相结合，来解决跟踪过程中可能会出现的目 标局部遮挡情况下与目标丢失的跟踪、目标重新捕获等问题；然后，建立一个云 台控制系统，该系统接收图像处理后的目标位置信息，并利用 Visual C++提供的 串口控件 MSComm 向云台发送相应的控制指令，实现实时监控跟踪目标；最后，构 建一个基于相关跟踪算法的实时目标跟踪系统实验验证平台，对本文改进的相关 跟踪算法进行验证。关键词：相关跟踪图像匹配目标跟踪电视跟踪AastractWith the swift development of science technology, the correlation tracking technology is widely used in the field of military tracking, Monitoring System, robot and so on. So the research of the correlation tracking technology has wide applications and important academic significance in the future. During the process of target tracking in practice, especially under the complex background, the target is very flexible, so it is more difficult to track target. The correlation tracking algorithm is the common one used under this condition. The correlation tracking algorithm is easy to be realized and has higher tracking precision. However, it is easy to cumulate the errors and has large computational complexity, which is not good for correlation tracking algorithm to be used in the actual tracking system. In allusion to the problems the correlation tracking algorithm has, some analyses and researches on the correlation tracking algorithm are made in this paper.It mainly makes some researches on the correlation tracking algorithm based on the image gray in the paper, and then build an experiment platform to validate the real-time target tracking system which realizes the real-time tracking target under complex background. First, we analyze and compare some correlation tracking algorithms which are usually used. Due to the results of the analysis and comparison, the mean removal c