摘 要

人脸识别系统的研究从20世纪60年代开始进行，随着计算机技术的发展对人脸识别的研究也不断深入。目前人脸识别技术已经可以应用在生活中的各个方面，给人类的生活带来大量的便利。人脸识别系统主要包括人脸图像检测及采集和人脸图像特征提取这两部分。人脸图像检测及采集即是对任意一张图片，通过某些方法对这张图片里面是否存在人脸进行搜索判断，如果存在人脸则返回人脸的相关信息。人脸图像特征提取就是针对人脸的某些特征进行提取，提取出每张人脸的特征以为后面的人脸匹配以及识别做准备。

目前，通过OpenCV来进行人脸图像检测和采集十分普遍。通过对大量人脸图像进行分析，得到人脸图像的相关特征，建立起人脸的模型，最后把输入的图片与建立起来的人脸模型相匹配，如果匹配成功，则说明这张图片里面存在人脸，则返回出此区域的人脸，若匹配失败，则此图片不含有人脸，输出空。

人脸识别算法主要包含如下算法，基于模板匹配的方法、基于奇异值特征的方法、子空间分析法、主成分分析法（PCA）、特征脸法等。通过对其中典型的人脸识别算法PCA、LDA、LBP的原理及特点的研究和分析，本文决定通过PCA（Principal Components Analysis）算法来对人脸图片进行特征提取。

本系统在人脸检测和采集方面具有较好的实现能力，在人脸特征提取和识别方面仍有一定的局限，具体表现为不同人脸图片之间的欧氏距离可能会出现更小的情况，仍然需要进一步加深自己的研究，完善系统。

关键词：人脸识别；人脸检测；人脸特征提取；主成分分析

Abstract

The research of face recognition system began in the 1960s. With the development of computer technology, the research on face recognition has been deepened. At present, face recognition technology can be applied to all aspects of life, bringing a lot of convenience to human life. The face recognition system mainly includes face image detection and acquisition and face image feature extraction. Face image detection and acquisition is to search for face picture in any picture. Some methods are used to search for the presence of face in this picture. If there is a face, the relevant information of the face is returned. Face image feature extraction is to extract some features of the face, extract the features of each face to prepare for subseqent face matching and recognition.

At present, face image detection and acquisition through OpenCV is very common. By analyzing a large number of face images, the relevant features of the face image are obtained, the model of the face is established, and finally the input image is matched with the established face model. If the matching is successful, it indicates that the image exists. then the face of this area is returned. If the match fails, the picture does not contain a face and the output is empty.

The face recognition algorithm mainly includes the following algorithms, based on template matching method, singular value feature based method, subspace analysis method, principal component analysis (PCA), feature face method and so on. Through the research and analysis of the typical face recognition algorithms PCA, LDA and LBP, this paper decides to extract the features of face images by PCA (Principal Components Analysis) algorithm.

The system has good implementation ability in face detection and acquisition, and there are still some limitations in face feature extraction and recognition. The specific expression is that the Euclidean distance between different face images may appear smaller. Still need to further deepen their research and improve the system.

Keywords: face recognition, face detection, face feature extraction,

PCA (Principal Components Analysis)

目录

第1章 绪论 1

1.1 研究意义 1

1.2 国内外研究现状 1

1.3 章节安排 2

第2章 系统环境搭建 3

2.1 Android简介 3

2.2 Android Studio简介 3

2.3 OpenCV简介 4

2.4 系统整体环境搭建 4

2.4.1 Android Studio安装 4

2.4.2 OpenCV安装 5

第3章 人脸特征提取算法 8

3.1 PCA（Principal Components Analysis） 8

3.2 LBP(Local Binary Patterns) 9

3.3 LDA(Linear Discriminant Analysis) 10

3.4 算法比较和选择 10

第4章 人脸识别系统搭建 11

4.1 通过OpenCV实现人脸检测 11

4.2 PCA算法实现人脸特征提取及识别 15

4.3人脸识别相关实现 15

第5章 系统整体测试与分析 18

5.1 人脸检测结果 18

5.2 人脸特征提取过程记录及结果分析 20

第6章 总结与展望 22

致谢 23

参考文献 24

附录 PCA算法相关代码 25