机器视觉的目标跟踪方法设计
摘 要
当前社会是一个高速发展的社会,经济在不断的发展，硬件水平也在不断提高。
随着人们生活现代化程度的提高，物联网机器视觉系统的应用也变得无处不在。
通过运用计算机图像处理技术，进一步提升物联网机器视觉系统的自动化水平,
尽可能地淘汰人工控制，是物联网机器视觉系统的重要研究方向。
物联网机器视觉离不开目标检测与跟踪，其中目标跟踪运用机器模拟人类的
眼睛对现实生活中的各种信息进行捕捉，并且对其跟踪定位，是机器视觉的关键
性技术。对于频监控中指定区域内的运动目标检测方法也因此而生。而目标检测
是目标跟踪的前提和基础，在日常应用中也被广泛应用。
文本通过分析图像分割和目标轮廓提取、meanshift 和粒子滤波的一些基本
概念和算法，阐述了目标检测与跟踪的方法与过程，经过单高斯背景建模、目标
检测这样的两个步骤进行，最后通过 MATLAB 对系统进行了仿真实验。
关键字：运动目标检测；目标跟踪；meanshift；粒子滤波
[abstract] The current society is a high-speed development of the society,
the economy in the continuous development, the hardware level is also
constantly improving.With the improvement of the modernization of
people's life, the application of machine vision system in the Internet
of Things has become ubiquitous.It is an important research direction of
the machine vision system of the Internet of Things to further improve
the automation level of the machine vision system of the Internet of Things
by using computer image processing technology and eliminate manual
control as much as possible.
The machine vision of the Internet of Things is inseparable from target
detection and tracking, in which target tracking uses machines to simulate
human eyes to capture all kinds of information in real life and track and
locate it, which is the key technology of machine vision.For frequency
monitoring in the designated area of the moving target detection method
is therefore born.Target detection is the premise and foundation of target
tracking and is also widely used in daily application.
By analyzing some basic concepts and algorithms of image segmentation and
target contour extraction, Meanshift and particle filtering, the paper
describes the method and process of target detection and tracking, through
two steps of single Gaussian background modeling and target detection.
Finally, the system is simulated by MATLAB.
Key word: toll; Image segmentation; Target extracted; MATLAB
