摘要视频中运动目标的检测识别是计算机视觉研究的主要问题之一,运动目标的检测以及运动分析在当前的研究领域十分活跃,其在解决智能视频监控,人机交互,智能交通系统等领域有着广泛而重要的应用,前景乐观。本文用高斯混合模型对运动目标进行检测提取,然后通过基于头肩轮廓的特征提取方法进行人体识别,从而完成视频中运动目标的人体识别方法研究。本文着重利用MATLAB 2012b开发平台,对视频文件或者图像文件进行仿真处理,然后观察结果。本文将重点介绍检测提取的算法和识别部分,并给识别部分的出仿真结果,然后加以详细描述。本文所用的混合高斯背景的更新技术可以随着背景信息的变化不断更新高斯模型,在一定程度的复杂背景下也能够较好的实现目标检测,是一种实现简单,运算量小的方法。30636
关键词 运动目标检测 人体识别 混合高斯模型 MATLAB仿真
毕业论文设计说明书外文摘要
Title The Research for Human Recognition of Moving Objects in Video
Abstract
Moving objects detection and recognition in video is one of the main resrearchs in computer vision. It has a broad and important applications in solving intelligent video surveillance, human-computer interaction, intelligent transportation systems and some other fields. In this paper, we introduce the way to detect moving objects extraction by using Gaussian mixture model, and then do the human recognition by characteristics of the head and shoulder contour , so that we can completing the research for human recognition of moving objects in video. This article will focus on the detection and identification section algorithm, and give out the simulation results by MATLAB 2012b. In this paper, Gaussian mixture background update technology can constantly update Gaussian model change with the change of background information ,a good objects detection can be achieved in some complicated background ,and it is a simple method to see the results with small amount of calculation
Keywords Moving objects detection Human recognition Gaussian mixture model MATLAB 2012b
目 次
1 引言 2
1.1 研究背景及意义 2
1.2 国内外研究现状 3
1.3 存在的问题 3
1.3.1视频运动目标检测的问题 3
1.3.2人体识别存在的问题 4
1.4 论文研究内容和结构 4
1.4.1 研究内容 4
1.4.2 论文结构 4
2.1 光流法 5
2.1.1 经典光流法的计算 5
2.1.2 经典光流法的目标检测 6
2.2 帧差法 7
2.3混合高斯背景模型检测 8
2.3 .1混合高斯模型介绍 8
2.3 .2混合高斯模型算法公式 9
3 人体识别方法概述 12
3.1 人体识别方法分类 12
3.2 基于头肩轮廓的特征提取方法 13
3.2.1 基本原理 13
3.2.2 进一步优化轮廓 14
3.2.3 BP神经网络训练 16
4 软件仿真实现 19
4.1 MATLAB R2012b软件介绍 19 MATLAB视频中运动目标的人体识别方法研究:http://www.youerw.com/tongxin/lunwen_26457.html