本科毕业设计--基于混合高斯建模方法的运动目标检测方法研究与实现.doc

摘 要 运动目标检测是计算机视觉研究领域的基础，它是从图像序列中去除静止的背景区域，将运动区域的前景检测并提取出来，如何有效地把感兴趣的目标如人、物等前景目标从复杂场景中分割出来，并对目标行为做出相应的检测，是计算机视觉研究的热点和难点。 本文对视频图像读取、图像灰度化等处理过程做了简单的阐述，并分别对均值滤波、中值滤波和高斯滤波这三种不同的滤波方法进行比较，分析各种方法的优缺点，同时提出使用高斯滤波方法。最后利用搭建在Visual C++6. 0上的OpenCV软件进行设计，使用混合高斯背景的建模方法实现对运动目标的检测。 在背景的各种建模方法中，混合高斯背景模型以其简便、灵活、高效的特点成为该领域的经典方法，能够很好消除背景的微小扰动对运动目标检测所产生的影响，比较准确地检测出复杂环境下的运动目标，具有较好的鲁棒性和实时性。 关键词: 运动目标检测；OpenCV；混合高斯背景模型 ABSTRACT Moving target detection is the foundation of computer vision research field, which is still to remove the background from an image sequence region, the prospect of a motion area detected and extracted, how effectively the target of interest, such as human, material and other foreground objects from a complex scene carved out, and make the appropriate test target behavior, is hot and difficult computer vision research. In this paper, the video image reading, image processing such as Gray did a brief explanation, and respectively, mean filter, median filter and Gaussian filter three different methods to compare the advantages and disadvantages of each method, At the same time proposed to use a Gaussian filter. Finally, build in Visual C ++ 6. OpenCV software 0 of design, using a mixed Gaussian background modeling methods to achieve the detection of moving targets. In the background of the various modeling methods, Gaussian mixture background model with its simple, flexible and efficient characteristics of a classical method in the field, can be a good influence to eliminate background small perturbations generated by the moving target detection, more accurate moving objects detected in complex environments, robust and real-time. Key Words: Moving Object Detection; OpenCV; The background model 目 录 1 绪论 1 1.1 研究背景和意义 1 1.2 国内外研究现状与发展趋势 1 1.2.1 国内外研究现状 1 1.2.2 研究的热点 3 1.3 研究内容 4 1.4 本章小结 4 2 视频图像检测预处理 5 2.1 视频读取 5 2.2 图像灰度化 6 2.4图像去噪 7 2.4.1 中值去噪 8 2.4.2 均值去噪 9 2.4.3 高斯去噪 10 2.5本章小结 11 3 运动目标检测算法研究 12 3.1 几种常见的运动目标检测方法的介绍 12 3.1.1