摘要计算机视觉的研究领域中，最热门的莫属视频目标的跟踪的研究了。随着科学技术的发展，国民安全意识的提高，安保智能监控系统、车辆飞行器等检测、制导、医用成像分析等领域对视频目标日益增长的需求，一定程度上推进了视频目标跟踪的研究与发展。视频目标跟踪的两大基本要求就是实时与准确。本文研究的就是视频序列的中的目标的跟踪。19019
本文以基于灰度的模板匹配为主，卡尔曼滤波预测为辅的方法，对动态复杂场景下的空中目标的进行了跟踪。主要的工作与研究成果如下：
（1）首先研究了运动目标检测的相关方法。采用了帧差法，对相邻的三帧图像作差，得到运动轮廓，用矩形包围轮廓，框出目标。检测出了目标大致位置。
（2）其次研究了目标模板图像的设计方法。采用了手动建立和自动建立两种方法。手动建立是指利用鼠标按键选定模板；自动建立则基于目标图像分割的理论，对二值化图像，抛弃了固定阈值，通过大津法确立合适阈值，寻找轮廓和外接矩形生成了比较合适的模板。
（3）最后研究了目标视频图像序列的跟踪方法。在模板匹配和基于卡尔曼的模板匹配的大前提下，分别采用了不同的模板的更新策略对目标进行跟踪，在一定程度上完成了对目标的准确跟踪。
实验表明，本文提出的跟踪方法在实时性和准确性方面基本达到了预设的指标。适用于复杂背景下，运动速度适中的图像目标。
关键词 复杂场景 二值化 大津法 模板匹配 卡尔曼滤波
毕业设计说明书（毕业论文）外文摘要
Title A Template Matching Method for Aerial Object Video Image Tracking
Abstract
Visual target tracking must be the most popular research field of computer vision. With the fast development of science and technology and the improvement of people’s safety awareness ,the demand has ever increasing in the field of intelligent monitoring system, vehicle detection , guidance ,medical analysis of imaging etc, which, to some degree , promote the development of visual target tracking . Accuracy and good real-time performance are the basic two requirements of Visual target tracking. In this paper, an example of visual target tracking has been studied.
In this paper , a method of matching template based on gray scale combined with Kalman filter prediction has been proposed to achieve good performance in an example of visual aerial target tracking under complex scene.
Firstly, several methods of target detection have been introduced .The method of target detection used in this paper is frame difference method which makes difference to three adjacent frames to get the contour, and mark the target by a rectangular contour enclosing. The approximate position of the target has been detected.
Secondly, we studied the methods of establishing a template image. The template can be established manually or automatically. The former establishes template by mouse-click, the latter one establishes template based on the theory of image segmentation by finding the contours in binary images and surrounding it with a rectangle. In addition, we give up the
method of a fixed threshold . Instead, a method called Otsu has been applied
毕业设计说明书（论文）外文摘要
to the setting of threshold for a binary image in this paper .As a result, a template image has been established . 空中目标视频跟踪模板匹配技术研究:http://www.751com.cn/zidonghua/lunwen_40174.html