论文总字数：27817字

摘 要

课堂点名是促进课堂教学，提高教学质量的关键环节，传统的人工点名模式存在点名效率低、替代点名等问题。人脸识别技术是随着人工智能技术发展起来的一种实用技术，在犯罪侦查、门禁系统等领域有着广泛而有效的应用。将人脸识别技术应用到课堂点名系统中能够有效解决目前人工点名模式存在的问题与缺陷，提高点名效率，并能防止误点、代替点名等问题。本次毕业设计利用python语言编程实现了一个基于人脸识别技术的点名系统，用典型的人脸识别算法对图片中的人脸进行识别，提取学生信息完成点名，提高课堂的效率。

本系统主要功能模块包括图像处理算法模块、人脸采集模块、人脸识别模块和点名模块。图像处理模块用于学生的人脸采集，将采集后的人脸进行预处理，并将处理后的人脸进行训练得到样本文件，其中记录了学生的人脸特征，在识别系统载入样本文件对学生进行人脸识别。点名系统管理模块主要负责人员的信息、点名信息的查询和管理。人脸采集模块用于样本文件和人脸识别过程中人脸的提取，在检测出人脸区域后将区域中的人脸经过人脸预处理后保存为图片，供训练时使用。人脸识别模块是将人脸与训练文件匹配，当置信度大于一定阈值时返回与人脸匹配的学号。点名模块将识别过程中返回的学号与数据库进行匹配，获得学生数据。

构建班级学生的人脸库，用典型的人脸识别算法对所拍摄的人脸进行识别，完成比对，生成图片中学生的数据，保存到点名库中。支持离线比对，并完成点名库的生成，并支持对没有识别的人脸进行手动点名；应尝试在线点名的实现，将点名信息实时的挂在图片上。

根据初在本次的课题完成任务中，需要完成如下工作：

首先，我通过将通过文献分析法研究人脸识别和课堂考勤的概念，以两个概念为基础对基于人脸识别的课堂点名系统进行概念界定。

其次，对基于人脸识别的课堂点名系统展开设计研究。在此阶段，我在相关点名设计模式的指导下，依据课堂点名系统的执行流程提出基于人脸识别的课堂点名系统的设计模式。

再次，进行基于人脸识别的课堂点名系统的功能设计，分为人脸采集、人脸识别、点名等。总结基于人脸识别的课堂点名系统的设计流程。

接着，对基于人脸识别的课堂点名系统进行整体构思与设想，为后续的技术开发工作提供方案。

最后，进入基于人脸识别的课堂点名系统的具体实施阶段，依据理论基础、设计模式对基于人脸识别的课堂点名系统进行技术的选择与开发工作。

关键词：人脸识别；点名系统；深度学习；可视化

Design and implementation of classroom roll call system based on face recognition

Abstract

The classroom call is the key link to promote classroom teaching and improve the quality of teaching. The traditional manual call out mode has the problems of low efficiency and substitution of naming. Face recognition technology is a practical technology developed with the development of artificial intelligence technology, and it has a wide and effective application in the fields of crime investigation and access control system. The application of face recognition technology to the classroom call system can effectively solve the problems and defects of the current manual call pattern, improve the efficiency of the call, and prevent the problem of wrong points and replace the call. This graduation design uses Python language to program a naming system based on face recognition technology. It uses typical face recognition algorithm to recognize the face in the picture, extracts the student information to complete the naming, and improves the efficiency of the classroom.

The main functional modules of this system include image processing algorithm module, face acquisition module, face recognition module and roll call module. The image processing module is used to collect students" faces. The collected faces are preprocessed, and the processed faces are trained to obtain sample files, in which the students" face features are recorded, and the sample files are loaded into the recognition system to recognize students" faces. The management module of the roll call system is mainly responsible for the information of the personnel, the inquiry and management of the roll call information. Face acquisition module is used for sample files and face extraction in the process of face recognition. After detecting the face region, the face in the region will be preprocessed and saved as a picture for training. The face recognition module is to match the face with the training file, and when the confidence is greater than a certain threshold value, the student number matching with the face is returned. The roll call module matches the student number returned in the identification process with the database to obtain the student data.

Build the class students" face library, use the typical face recognition algorithm to recognize the face taken, complete the comparison, generate the picture of students" data, save to the roll call library. Offline comparison is supported, and the generation of roll call library is completed, and manual roll call is supported for faces that have not been recognized. Try to implement the online roll call, hang the roll call information on the picture in real time.

According to the initial task of this topic, the following work needs to be completed:

First of all, I will study the concepts of face recognition and class attendance through literature analysis, and define the concept of class attendance system based on face recognition based on the two concepts.

Secondly, the class roll call system based on face recognition is designed and researched. At this stage, I put forward the design mode of class call system based on face recognition according to the implementation process of class call system under the guidance of relevant call design mode.

Again, the function design of class roll call system based on face recognition is divided into face acquisition, face recognition, roll call and so on. Summarize the design process of class roll call system based on face recognition.

Then, the class roll call system based on face recognition is conceived and envisaged as a whole, to provide a plan for the subsequent technology development.

Finally, it enters the specific implementation stage of the class roll call system based on face recognition, and selects and develops the technology of the class roll call system based on face recognition according to the theoretical basis and design pattern.

Key words: Face recognition; Roll call system；Deep learning;Visualization

目 录

摘 要 I

Abstract II

第一章　引 言 5

1.1 课题研究背景与意义 5

1.1.1 研究背景 5

1.1.2 研究目标 5

1.2 人脸识别技术发展现状 6

1.3 主要工作内容 6

1.4 论文框架 6

第二章　系统技术介绍及分析 7

2.1 人脸关键点对齐主要算法 8

2.1.1 人脸关键点定位 8

2.1.2 人脸对齐 8

2.1.3 级联回归方法 8

2.1.4 CPR(Cascaded Pose Regression) 8

2.1.5 ESR(Explicit Shape Regression) 8

2.1.6 ERT(Ensemble of Regression Trees) 9

2.2 Face Recognition核心函数 9

2.2.1 人脸检测 9

2.2.2 检测面部特征点 9

2.2.3 面部编码 9

2.2.4 编码中实现准确识别 10

2.3 本章小结 10

第三章　需求分析 11

3.1 系统功能需求分析 11

3.2 系统非功能需求分析 11

3.2.1 系统可维护性与扩展性 12

3.2.2 系统安全性与保密性 12

3.2.1 系统易用性 12

3.3 本章小结 12

第四章　基于人脸识别的课堂点名系统实现 13

剩余内容已隐藏，请支付后下载全文，论文总字数：27817字