Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx

Contents lists available at ScienceDirect

Journal of King Saud University – Computer and Information Sciences

journal homepage: www. sciencedirect.com

An android based course attendance system using face recognition

Dwi Sunaryono ^a, Joko Siswantoro ^b,uArr;, Radityo Anggoro ^a

a Department of Informatics, Institut Teknologi Sepuluh Nopember, Surabaya, Indoneisa

b Department of Informatics Engineering, Universitas Surabaya Jl. Kali Rungkut, Surabaya 60293, Indonesia

a r t i c l e i n f o

Article history:

Received 11 September 2018

Revised 28 November 2018

Accepted 12 January 2019 Available online xxxx

Keywords:

Course attendance system Face recognition

Android based Smartphone

a b s t r a c t

Student attendance system is needed to measure student participation in a course. Several automated attendance systems have been proposed based on biometric recognition, barcode, QR code, and near field communication mobile device. However, the previous systems are inefficient in term of processing time and low in accuracy. This paper aims to propose an Android based course attendance system using face recognition. To ensure the student attend in the course, QR code contained the course information was generated and displayed at the front of classroom. The student only needed to capture his/her face image and displayed QR code using his/her smartphone. The image was then sent to server for attendance pro- cess. The experimental result shows that the proposed attendance system achieved face recognition accu- racy of 97.29 by using linear discriminant analysis and only needed 0.000096s to recognize a face image in the server.

copy; 2019 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an

open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Student attendance is an important factor for students to suc- ceed in a course. In a certain university, student attendance in a course is also used as one of requirements for student to take the exam (Islam et al., 2017). A conventional approach to record stu- dent attendance is performed by asking every student to sign on an attendance list that passes through all students during the beginning of lectures. However, this approach is inefficient in term of time and can potentially lead to a fraud especially in a large class, where a student can sign on the attendance list for other stu- dents who are not present in the class. To avoid the happening of fraud, sometimes the lecturer calls out the names of students who have signed on the attendance list one by one. This method will take the lecture time and will have an impact on the effective- ness of lecture (Mohamed and Raghu, 2012). A modern approach to record attendance is by using automated attendance system. Sev- eral automated attendance system have been proposed by employ-

uArr; Corresponding author.

E-mail addresses: dwis@if.its.ac.id (D. Sunaryono), joko_siswantoro@staff.ubaya. ac.id (J. Siswantoro), onggo@if.its.ac.id (R. Anggoro).

Peer review under responsibility of King Saud University.

Production and hosting by Elsevier

ing biometric recognition, such as fingerprint recognition (Mohamed and Raghu, 2012; Rao and Satoa, 2013; Soewito et al., 2015; Zainal et al., 2016; Zainal et al., 2014), face recognition (Chintalapati and Raghunadh, 2013; Fuzail et al., 2014; Mehta and Tomar, 2016; Raghuwanshi and Swami, 2017; Sayeed et al., 2017; Wagh et al., 2015; Wati Mohamad Yusof et al., 2018) and palm vein recognition (bayoumi et al., 2015) to recognize students who are present and record their attendance. The other proposed attendance systems used barcode (Noor et al., 2015), QR code (Rahni et al., 2015), RFID (Arulogun et al., 2013; Bhalla et al., 2013; Hussain et al., 2014; Rjeib et al., 2018) and near field com- munication (NFC) mobile device (Mohandes, 2017) to obtain stu- dent ID for attendance process. Some attendance systems were developed in portable device (Mohamed and Raghu, 2012; Zainal et al., 2016, 2014) and smartphone (Islam, et al., 2017; Mohandes, 2017; Noor et al., 2015; Rahni et al., 2015; Soewito et al., 2015).

Rao and Satoa (2013) have proposed an employee attendance management system using fingerprint recognition. Every check in and check out times, employees needed to scan their fingerprint to record attendance. Minutiae-based matching c

剩余内容已隐藏，支付完成后下载完整资料

基于Android的课程考勤系统，使用人脸识别

1 介绍

学生出勤是学生在课程中取得成功的重要因素。在某所大学，学生参加课程也被视为学生参加考试的要求之一（Islam et al.,2017）。记录学生出勤率的传统方法是通过要求每个学生在讲座开始时签署通过所有学生的出勤名单来进行。然而，这种方法在时间上是低效的，并且可能潜在地导致欺诈，尤其是在大班级中，其中学生可以在出席名单中为在班级中不存在的其他学生签名。为了避免欺诈行为的发生，有时讲师会逐一列出在出勤名单上签名的学生的姓名。这种方法将占用讲座时间，并将对讲座的有效性产生影响。一种现代的记录出勤率的方法是使用自动考勤系统。通过采用生物识别，例如指纹识别，人脸识别和手掌静脉识别来识别在场的学生并记录他们的出勤率。其他提议的考勤系统使用条形码，QR码，RFID和近场通信（NFC）移动设备获得出勤过程的学生ID。一些考勤系统是用便携式设备和智能手机提出了一种使用指纹识别的员工考勤管理系统。每次办理入住和退房手续时，员工都需要扫描指纹来记录出勤情况。基于细节的匹配与基于对齐的贪婪匹配相结合用于识别所提出的考勤系统中的扫描指纹。虽然作者报告说所提出的系统易于使用且成本低，但所提出的系统不适合课程考勤系统，因为如果同时存在大量课程则系统需要大量的指纹记录设备。而且，如果课程中有大量的学生，那么系统将导致长时间且耗时的排队。Rao，Satoa和Zainal等，提出了一种基于指纹识别的学生考勤系统便携式设备。建议的系统要求学生将他/她的指纹扫描到设备上以进行考勤。出勤数据仅存储在设备上。设备没有直接连接到服务器，因此讲师需要在课时后手动将数据备份到服务器。此外，如果同时有许多课程，则需要许多设备。Soewito，et al。（2015）在Android智能手机上提出了一个使用指纹和GPS与支付系统集成的员工考勤系统。从用户智能手机，系统记录指纹，考勤时间和智能手机上可用的GPS位置坐标，以避免排长队和假出勤。但并非所有Android智能手机都配备了指纹扫描仪。此外，在Android智能手机上通过GPS记录用户位置是不准确的。根据Bauer（2013）的说法，Android智能手机上的GPS偏离了实际位置大约10-93米。因此，在办公室外面但仍然离办公室足够近的员工可以记录在场。基于指纹识别的几乎所有提出的考勤系统都没有报告识别准确性，除了Zainal等人提出的系统外。，这是85％，27名学生的总识别时间约为7-9分钟。而且，基于指纹识别的考勤系统存在缺陷。正如Zainal等人报道的那样。如果指纹是潮湿的，有脏污或破损的，则系统无法识别。

Chintalapati和Raghunadh（2013），Fuzail等人（2014）Mehta和Tomar（2016），Raghuwanshi和Swami（2017），Sayeed等（2017），Wagh等（2015），Wati Mohamad Yusof等人，2018提出了基于人脸识别的自动化学生考勤系统。提出的系统使用相机同时捕获所有学生的面孔或逐个使用主成分分析（PCA）和局部二元模式（LBP）结合一些分类器进行人脸识别，并通过使用LBP和欧几里德距离为80名学生实现了78％的最佳分类准确率。Sayeed等人（2017）提出了使用PCA和欧几里德距离进行考勤系统的实时人脸识别。Wati Mohamad Yusof等（2018）提出了一种使用面部识别的基于互联网的实时考勤系统。所提出的系统采用Haar级联进行人脸检测，结合LBP进行人脸识别。Chintalapati和Raghunadh所提出的系统效率不高，因为他们只用摄像头通过一个接一个捕捉学生的面部图像。Fuzail等人（2014年）。Wagh等人（2015年），Mehta和Tomar（2016年）以及Raghuwanshi和Swami（2017年）使用相机一次性捕捉所有在教室里的学生的面孔。该策略可以避免在考勤过程中出现排队。然而，使用这种策略的考勤系统在人脸识别方面的准确性较低，如Raghuwanshi和Swami（2017）所报道，使用主成分分析（PCA）和欧几里德距离和线性判别分析（LDA）分别为53.33％和60％。

条形码的使用，QR码，RFID和NFC是在考勤系统中记录学生身份的另一种选择。Arulogun等人，2013，Bhalla等人，2013，Hussain等人，2014，Noor等人，2015，Rahni ，Rjeib等人提出的系统中的出勤过程非常简单，学生只需要使用系统扫描他们的学生卡中包含条形码，QR码或RFID来记录出勤率。在基于NFC的考勤系统（