基于Android的人脸识别--论文Word文件下载.docx

1、物联网工程学 院 计算机科学与技术专 业学 号学生姓名指导教师I随着科技的发展和进步，社会进入一个信息自动化的时代.对于信息的安全性，和数据的保密性，尤为重要，在当代信息就是一切的基础要素.每个人所获得信息都不同，所有信息需要对应生物识别，因此基于人体生物特征的识别软件的需求应运而生.在人体的多种生物特征中，人脸识别具有普遍适用性，方便快捷性，整个识别过程可以在目标无意识下进行.从而达到信息的自动化处理，又需要保证信息的安全性.人脸识别技术可以应用在识别人类身份方面应的各个领域，它提取的人类的生物特征作为密钥，具有高度安全性，和密码唯一性的特征.随着技术的不断发展，在 android 设备可以

2、采用人脸识别技术以完成工作.面部识别作为一种生物识别方法具有非接触的特点，安全方便.它广泛应用于人机交互，交易认证，安全等领域.近年来随着移动互联网和嵌入式电脑的发展，在嵌入式系统上运行人脸识别，这类应用程序在远程支付和个人信息安全方面具有巨大的潜力.该应用程序是在 Android 操作系统的上运行的，该系统基于Linux 内核，作为当前使用最广泛的智能设备.面部识别的过程包括面部检测，面部归一化和人脸识别.本文对其中图像预处理和人脸识别部分做了详细的研究和介绍.本系统中所采用了主成份分析（PCA）方法进行人脸特征取,对其中主要的原理和步骤进行着重分析.并成功开发了一个具有良好的识别率应用程序

3、.人脸库，选取 20 张图片构成人脸库，每张图都是经过标准处理的，读入人脸库，并做降维处理，二维向量转换为一维向量，则训练集是一个 36000*20 的矩阵，待测图片也为标准处理的 180200 像素，即待测集为一个 36000*1 的矩阵.使用脸部检测算法并在目标上找出人脸图像，对图像进行预处理，然后再进行脸部识别.图片预处理是准备识别工作的一部分的部分.使用减少图像大小，降低噪音和放大图像对比度等方式，它能够最大化减少计算量.适应移动设备的使用.系统主要分为 2 个模块，系统中的输入模块是人脸采集和信息输入过程以获取训练图像，作为参考的样本保存到数据库.识别模块是系统对人脸进行比对识别身份

4、的一个过程.把数据库中的训练图像与测试图像相比较的并识别，然后识别输出结果.文章的主要内容如下：1） 讨论面部检测方法.使用 Adaboost 算法和 Haar 检测人脸的功能.2） 研究图像预处理方法.标准化图像以便最小化存储空间并加快计算速度.3） 综合各种面部识别算法，基于本应用使用的 PCA 算法进行讨论分析.4） 完成从脸部检测到人脸识别的所有功能.充分验证程序的有效性.讨论结果和识别策略.关键词：人脸识别；Android 应用；PCA 算法；OpenCV 库.IIIABSTRACTWith the development and progress of science and te

5、chnology, it is especially important for the security of information and the confidentiality of data due to the society has entered an era of information automation. Information is the basic essential factor of everything in the present state. And all the information needs to correspond to biologica

6、l identification because each person gets different information,therefore，the demand for identification software of basing on human biological characteristics is begot in response to the needs of the times. Among various biological features of human body, the face recognition is universal, convenien

7、t andfast, and the whole recognition process can be performed under the targets unconsciousness to achieve the automation processing of information, and ensure the safety of information.Face recognition technology also can be applied to various fields of human identification; it extracts human biome

8、tric features as keys with high safety and the uniqueness of the password. With the continuous development of technology, the face recognition technology can be used to complete the work in the Android device. Face recognition is a biometric identification method; its characteristics are non-contact

9、, safe and convenient and are widely used in human-computer interaction, transaction authentication, security and other fields. With the rapid development of mobile Internet and embedded computer in recent years, it works face recognition in the embedded system, and there are great potential for the

10、 remote payment and personal information security in such applications. This application runs on the Android operating system, based on the Linux kernel, and as the most widely used smart device today, the process of face recognition includes face detection, facial normalization and face recognition

11、. This paper made a detailed study and introduction for the image preprocessing and face recognition. This system adopts principal component analysis （PCA） method to sample for face feature, and mainly analyze the principle and the main steps to successfully develop an application with good recognit

12、ion rate.Face database, was formed by chosen 20 pictures. Each picture is processed bystandard, and is read into the face database and done by dimensionality reduction.If the two-dimensional vector is transformed into a one-dimensional vector, the train set is a matrix of 36000*20, and the picture t

13、o be tested is also 180 x 200 pixels ofstandard processing, the test set is a matrix of 36000*1. The face detection algorithm is used to find the face pictures on the target to preprocess the pictures and then perform face recognition. Pictures preprocessing is part of the process of preparing recog

14、nition to reduce the pictures size, reduce noise, and amplify pictures contrast, which can reduce the amount of calculation to maximization to adapt the use of mobile devices.The system is mainly divided into 2 modules. The input module is the face acquisition and information input process to get tr

15、aining picture and save them into database as reference sample. The identification module is processes of the system distinguish identity by comparison, which compare and distinguish the training picture of database with test picture and output result of identification.Keywords: face recognition, Android app, PCA algorithm, OpenCV library.目录第 1 章绪论11.1 引言11.2 人脸识别背景介绍11.3 人脸识别应用场景11.4 人脸识别的特点：21.5 人脸识别技术研究现状21.6 人脸识别的过程31.7 选择 Android 平台的优势41.8 本章小结4第 2 章技术支持52.1 Android 系统框架52.2 Activity 的生命周期62.3 OpenCV 库72.4