基于OpenCV的人脸检测
摘 要
人脸检测是指使用计算机在动态的场景或复杂的背景中检测出人脸的存在，并确定所有人脸的位置与大小的过程。人脸检测技术不仅是人脸识别、表情识别、人脸跟踪等技术的重要前提条件，同时在模式识别、人机交互、智能监控、视频检索等领域也引起了广泛的重视。
本论文简单介绍了国内外人脸识别技术研究及应用的发展现状及其难点分析。在第二章中介绍了包括基于知识的方法、特征不变量方法、模板匹配方法、基于外观的方法四类检测方法；在第三章简单介绍了四种经典的检测方法，包括特征脸、神经网络、隐马尔可夫模型方法、支持向量机。
在第四章重点分析了AdaBoost算法中集成机器学习的一个重要机制：多个弱分类器集成的方法，机器学习中的弱学习到强学习，集成的关键是投票，最简单的方法是“绝对多数”的方法。在第五章详细分析了AdaBoost算法检测速度快、可以检测任意尺度的图像的特点。
在这个理论基础上，本文中人脸算法的研究基于OpenCV开源代码，在OpenCV开源代码中设计了一些基础的数据类型和一些帮助数据类型。由于OpenCV的源代码完全开放，本文的研究中利用这套代码在PC上以Visual C++集成开发环境做平台搭建了一个基于OpenCV的人脸检测系统，进行了人脸检测仿真，并对其结果进行分析。
关键词：人脸检测；AdaBoost；分类器；OpenCV
Face Detection Based on OpenCV
Abstract
Human face detection means that for a given image or video，to determine whether it contains face regions，if so，determines the number, the exact location and the size of all the faces．Human face detection is not only a necessary precondition of face recognition，expression recognition technology, face tracking，but also，it plays ail important role in applications like in the intelligent human-computer interaction，video conferencing，intelligent surveillance，video retrieval and so on．Therefore，face detection technology attracted widespread attention in pattern recognition，computer vision，human-computer interaction and other fields．
This paper describes the national and international recognition technology applications research and development and analysis the difficult situation. In the second chapter to include Knowledge-based Methods, Feature Invariant Approaches, Template Matching Methods, Appearance-based Methods. In the third chapter describes the four classical detection methods, including Eigenface, Artificial Neural Network, Hidden Markov Model, Support Vector Machines.
In chapter 4 on the analysis of the AdaBoost algorithms integration of machinery to study of an important mechanism : more integrated approach in the classification of the study of study at the learning and integration is the key to the vote, the simplest method is the absolute majority . In chapter five detailed analysis AdaBoost algorithm test speed, you can detect any scale the image quality.
This paper’s research is based on the OpenCV source code， in which some basic data type and helping data type were created，and because of the opermess of the code，we build a human face detection system in the Visual C++ environment.
Keywords: face detection ; AdaBoost ; classifier ; openCV
2
目 录
摘 要 ........................................................................................................................................................... 1 Abstract .......................................................................................................................................................... 2 第1章 人脸检测 ........................................................................................................................................ 5
1.1 背景 .............................................................................................................................................. 5 1.2 目前的研究状况 .......................................................................................................................... 6 1.3 概念 .............................................................................................................................................. 6 1.4 人脸检测及其合成技术的应用领域 .......................................................................................... 6 1.5 人脸检测评价标准 ...................................................................................................................... 7 第2章 检测方法 ........................................................................................................................................ 9
2.1 基于知识的方法 .......................................................................................................................... 9 2.2 特征不变量方法 .......................................................................................................................... 9 2.3 模板匹配方法 ............................................................................................................................ 10 2.4 基于外观的方法 ........................................................................................................................ 11 第3章 经典方法概述 .............................................................................................................................. 13
3.1 特征脸 ........................................................................................................................................ 13 3.2 神经网络 .................................................................................................................................... 13 3.3 隐马尔可夫模型方法 ................................................................................................................ 14 3.4 支持向量机 ................................................................................................................................ 14 第4章 Adaboost算法 ............................................................................................................................. 16
4.1 概述 ............................................................................................................................................ 16
4.1.1 Adaboost算法简介 ....................................................................................................... 16 4.1.2 Adaboost人脸检测算法 ............................................................................................... 17 4.2 弱学习与强学习 ........................................................................................................................ 18 4.3 PAC 基本模型 .......................................................................................................................... 19
4.3.1 概述 ................................................................................................................................ 19 4.3.2 基本概念 ........................................................................................................................ 19 4.3.3 PAC模型的不足 ........................................................................................................... 20 4.4 Boosting 方法 ........................................................................................................................... 21 4.5 Adaboost算法性能分析 ........................................................................................................... 21 第5章 矩形特征与积分图 ...................................................................................................................... 22
5.1 引言 ............................................................................................................................................ 22 5.2 矩形特征 .................................................................................................................................... 22
5.2.1 概述 ................................................................................................................................ 22 5.2.2 特征模板 ........................................................................................................................ 23 5.3 积分图 ........................................................................................................................................ 24
5.3.1 积分图的概念 ................................................................................................................ 24 5.3.2 使用积分图计算 ............................................................................................................ 25 5.4 Haar特征值计算 ....................................................................................................................... 27 第6章 人脸检测的实现 .......................................................................................................................... 31
6.1 OpenCV概述 ............................................................................................................................... 31
6.1.1 OpenCV简介................................................................................................................. 31
3
6.1.2 应用领域 ........................................................................................................................ 31 6.1.3 OpenCV的起源............................................................................................................. 32 6.1.4 OpenCV的基本结构 ..................................................................................................... 32 6.1.5 OpenCV的特征 ............................................................................................................. 33 6.2 OpenCV在检测中的应用 ........................................................................................................... 34
6.2.1 编译OpenCV................................................................................................................. 34 6.2.1 为VC++ 2008 Express配置OpenCV环境 ................................................................. 36 6.3 实验结果 .................................................................................................................................... 37 6.4 结论： ........................................................................................................................................ 39 参考文献...................................................................................................................................................... 40 致 谢 ......................................................................................................................................................... 42 附录一 人脸检测源程序 .......................................................................................................................... 43 附录二 外文翻译 ........................................................................................................................................ 48
4
相关推荐：
loading