毕业设计说明书中文摘要智能监控[1]是利用视频分析技术来实现对无人值守监控场景进行智能的“监视”，对监控场景中的异常情况或监控对象的异常行为能及时进行检测、分析，克服了传统监控系统的事后性。在智能监控系统的研究中，人体的检测、识别和跟踪是重要内容之一，也是该领域最前沿的关注对象。本文研究的是基于静态背景下的运动人体识别，主要包括两个部分：运动人体检测、运动人体识别。78804

首先，本文介绍了图像预处理方面的知识，包括图像去噪、直方图灰度均衡化、图像分割。

其次，本文针对传统的运动目标检测方法，实现了一种基于边缘检测与三帧差分相乘相融合的运动目标检测算法。该算法较传统检测算法，能更好地减小噪声和光线等造成的影响，使获得的目标轮廓更加清晰。

最后，本文描述了基于人体头肩椭圆模板的运动人体识别算法。首先构造人体头肩椭圆模板，确定待识别头肩部起始位置坐标；接下来再进行级联匹配；最后根据设定的阈值进行人体目标识别。

毕业论文关键词  人体识别   人体检测  头肩模板  图像预处理

毕业设计说明书外文摘要

Title    The Research on Moving Human Recognition Method of Intelligent Monitoring         

Abstract

Intelligent Monitoring is able to achieve intelligent “surveillance” on unattended surveillance scenes with the video analysis technologies, making timely detection and analysis of abnormal conditions and behaviors in the monitoring scene, overcoming the afterwards nature of traditional monitoring systems。 Human body detection, identification and tracking is one of the major research contents of intelligent monitoring, which is also a forefront direction in the field of computer vision in recent years。 This study is about moving body recognition under the static background, which mainly consists of two parts, namely moving body detection and moving body identification。

Firstly, the paper introduces the knowledge of image preprocessing, including image denoising, histogram gradation equalization and image segmentation。 

Secondly, this paper puts forward moving object detection algorithm based on the multiply and integration of edge detection and three-image difference against the traditional moving object detection method。 Compared with the traditional detection algorithms, this algorithm can reduce the impact of noise and light better, to make the target profile clearer。 

Finally, this paper proposes moving body recognition algorithm based on human head and shoulders ellipse template。 First, it constructs human head and shoulders ellipse template to determine the coordinate of the initial position of the head and shoulder; then it makes cascade matching; finally it makes human object recognition based on the set threshold。

Keywords  human body recognition   human body detection   head and shoulders template   image preprocessing

目   次

1  引言  1

1。1研究目的及意义  1

1。2国内外发展及现状  1

1。3本文研究内容及组织结构  2

2  图像预处理  3

2。1  图像去噪  4

2。2  直方图灰度均衡化  4

2。3  图像分割  4

2。4  本章小结  6

3  运动人体检测  7

3。1  传统运动人体检测方法  7

3。2  改进运动人体检测方法  8

3。3  本章小结 11

4  运动人体识别 12