摘  要：现如今随着科学技术的不断发展，人们的个人信息已经无法得到有效的保密，人脸的识别便走入了研发者和大众的视线中，因其在安全验证，档案管理，犯罪识别，人机交互等领域的普及应用，也逐渐深入广大群众的生活中去，成为了模式识别的一个新的研究方向和重点。

本文基于OpenCV来研究人脸的识别和跟踪功能，采用Adaboost算法和Haar特征。论文介绍人脸识别在生活中的应用与前景；还介绍了此次人脸识别研究的环境：OpenCV，经过多次比较研究得到的最适合本次设计的算法：Adaboost算法介绍和其历史；接着是算法在此次研究中的作用和影响，详细分析算法；最后展现本次研究的成果，以及对结果的更深层次的想法。

本次设计在剑桥AT&T实验室的ORL人脸库的基础上，研究使用SVM训练进行人脸识别的成功率；在此基础上深入研究动态情况下，此算法识别人脸的成功率。实验最终证明Adaboost算法和SVM训练能够保证人脸检测识别的成功率，在这个领域有着很大的前景和未来。

关键词：人脸识别  OpenCV  Adaboost算法  SVM训练

The Face Recognition System based on Adaboost Algorithm in OpenCV

Abstract：Nowadays with the development of science and technology, people's personal information has been unable to get effective confidentiality, so face recognition has entered the developers and the public's sight. Because of its popularization and application in the field of Security verification, file management, crime identification, human-computer interaction, It also gradually penetrated the masses' lives and Become a new research direction and focus of pattern recognition.

This article is based on OpenCV to study the recognition and tracking of faces, using Adaboost algorithm and Haar characteristics. First, the paper introduces the application and prospect of face recognition in life. The second introduction is the context of the face recognition study: OpenCV。The algorithm that is best suited for this design is the Adaboost algorithm and its history; Then introduce the function and influence of the algorithm in this study, the detailed analysis algorithm; Finally, the results of this study and the deeper thoughts on the results are presented.

The design is based on the ORL face library at AT&T laboratories in Cambridge, and studies the success rate of face recognition using SVM training. On the basis of this, we will study the success rate of identifying faces  under the dynamic situation with this algorithm. The experiment finally proves that the Adaboost algorithm and SVM training can guarantee the success rate of face detection recognition, and they have a lot of foreground and future in this area.

Key Words: Face recognition  OpenCV   Adaboost algorithm  SVM training

目  录

摘要Ⅰ

Abstract-Ⅱ

目录Ⅲ

图清单V

表清单V

1 绪论1

 1.1课题背景-1

 1.2研究目的与意义1

 1.3国内外发展概况2

    1.3.1国外发展情况-2

    1.3.2国内发展情况-2

2系统介绍与方案选择-4

 2.1系统介绍-4

    2.1.1模块与功能分析-4

    2.1.2运行环境分析-4

 2.2方案选择-5

2.2.1系统环境选择-5

2.2.2开发工具选择-6

2.2.3算法选择-6

3系统流程与算法设计-8

 3.1结构流程设计-8

3.1.1功能模块介绍-8

3.1.2结构设计-8

 3.2算法与特征详述-9

     3.2.1  Adaboost算法介绍-9

3.2.2  Haar-like特征介绍-10

 3.3算法与代码实现12

3.3.1图像预处理过程12

3.3.2利用Adaboost算法与Haar特征进行人脸检测14

3.3.3  SVM训练过程16

3.3.4视频跟踪分帧处理-17