Impersonation Detection in Online Examinations

Pooja Mahesh K Selvajyothi

Student, ECE – Design & Manufacturing Assistant Professor, ECE - Design & Manufacturing
IIITD&M - Kancheepuram IIITD&M - Kancheepuram
Chennai, India Chennai, India
[email protected] [email protected]

Abstract—In today’s world that witnesses an ever-changing Various proposals have been made to eliminate
scenario in the technical fields, the concept of an “examination” impersonation in online examinations. However, most of them
too has changed. From the traditional methods of pen-and-paper involve only a one step biometric verification [3], or verification
and Optical Mark Recognition (OMR), it has evolved into ‘Online of behavioral characteristics [4, 5], which are not always
Examinations’, which are a lot more flexible, time-saving as well reliable.
as require much fewer resources. However, they have certain
drawbacks too. While the uncertainties involved in the working of Various types of impersonation challenges that are faced in
the electronic equipment and ‘server break-down’ do affect the an online examination are discussed in [6]. In this paper, a
examination, the developments in technology have helped method is designed and implemented to eliminate these
overcome these. However, the most important problem faced by impersonation types. Here, a system of impersonation detection
online examinations is the authenticity of the student who is taking is proposed that consists of a two-step biometric verification to
it. Impersonation, as it is called, is in fact, due to the negligence (or completely eliminate the chances of impersonation. The first
in certain cases due to the cooperation) of the human factors that step is an already existing fingerprint verification, and the
are present at the examination centre. In this paper, the aim is to second step is a face detection and recognition done throughout
eliminate these human factors, to ensure that impersonation, if the duration of the examination, to avoid any form of
any, can be easily detected, and the impersonator is not allowed to impersonation.
take the examination. For this purpose, two- step biometric
verification of the candidate is done, one of which extends The paper is organized as follows. The proposed method of
throughout the duration of the examination. In this manner, avoiding impersonation with minimal human interference has
impersonation can be avoided. been discussed in Section II. In Section III, the implementation
methodology of the proposed system is discussed, followed by
Keywords—Impersonation, Online Examination, Face results and conclusions.
Recognition, Fingerprint Verification, Aadhaar Card.

I. INTRODUCTION
II. PROPOSED METHOD
Online examinations have slowly begun gaining popularity
and are taking over the traditional methods of pen-and –paper
A. Existing System of Online Examination
and OMR response sheets in India. Initially, however, it did face
numerous challenges. Back in the year 2009, when the Delhi The present system of online examinations includes a
Common Admission Test (DCAT) went online, the organizers registration phase where the details of the candidate are entered.
were forced to offer examinees another chance at answering the A passport size photograph, as well as a thumbprint is taken as
paper as there were computers getting crashed and numerous proofs of identity. In some cases, a signature, either a scanned
candidates were unable to complete the test [1]. copy or a digital signature is collected as well.
Almost a decade after the DCAT incident, online The next phase is that of examination. Only those candidates
examinations have become a trend. Most organizers prefer with a valid admit card are allowed into the examination hall.
online examinations over offline examinations due to factors The fingerprint verification of the candidate is done and a system
like cost effectiveness as well as time and energy conservation. is allotted to the candidate if the document verification yields a
Students too have begun adapting to the new method, though positive result.
gradually. The advantages of an online examination over the Now, the candidate can enter his/her registration number and
offline one, from a student’s perspective include the ease of a password on a screen which is later redirected to the page of
changing a selected answer, the flexibility to choose a date of examination.
examination as well as the timer set at the server, which helps
them manage their examination time effectively [2]. In some cases, the login page also shows the image of the
student to whom the specific system is allotted.

978-1-5386-3864-4/17/$31.00 © 2017 IEEE

 Fig. 1. Flowchart of the proposed impersonation detection method.

B. Proposed System of Online Examination

The major problem that is faced by the online examinations
is impersonation as already mentioned. The aim of this paper is
to introduce an online examination system that can tackle the
issue of impersonation with very little human factors involved.
This is necessary as the trustworthiness of the invigilators cannot
Fig. 2. Page showing instruction for the verification
always be guaranteed.
procedures.
There are two phases involved in the process of conducting
the online examination (Fig.1):  After fingerprint verification (during which an image of
the candidate is also captured), the screen moves onto the
 The Registration Phase, and login page, as shown in Fig. 3.
 The Examination Phase

C. The Registration Phase

At the time of Registration, the candidate is required to fill
in personal details like name, date of birth, gender and
educational qualifications to prove eligibility to take the
examination. The Aadhaar Card information is also collected
along with a passport sized photograph of the candidate.

D. The Examination Phase

The following are the sequence of events at the examination
center:
Fig. 3. Login page with photograph verification.
 The candidate must verify his/her examination center and
check for the system allotted.  The candidate has to verify if the photograph matches
 Upon entering the examination hall, on the monitor that with that of his/her own. If it matches, then the candidate
is allotted to each of the candidates, a list of instructions can enter the username and password and login to the
would be displayed, as shown in Fig. 2.
 page displaying instructions to take the examination (Fig. III. IMPLEMENTATION OF PROPOSED SYSTEM
4).
A. Webpage Development
 If the photograph is not matching, then the invigilator has
to be contacted. The front end of the system proposed has been developed
ensuring that it is user-friendly and simple to implement. Clear
 From the instructions page, the user/candidate is taken to instructions are given to the candidates in order to ensure that
the examination page, at a pre-set time. The clock would there is no ambiguity and discomfort to the candidates appearing
be set at the server. for the examination.
 Once the examination starts, the image of the student is The detailed front-end process is already discussed in
compared to that of what was captured while the Section II. The front-end is made user-friendly and easy to
fingerprint was entered. If the image matches, the comprehend. Also, the system is so designed that any external
examination continues. Else, after a set number of limits device (or another system), cannot be connected to it either in
of impersonation threat, the examination pauses, and an the wired, or the wireless form. This is done to avoid any
alert is given as shown in Fig. 5.Along with the possibility of wireless accessibility to the online examination by
examination being paused, a message will be sent to the an impersonator who is located elsewhere, while the actual
examination in-charge (or the higher authorities of the candidate is present before the original system in order to prove
body conducting the examination), warning them of an the candidature authenticity.
impersonation attempt (Fig. 6).
B. Fingerprint Recognition
 The candidate would not be able to resume the exam until
his/her candidature is verified by the concerned A simple optical fingerprint sensor is connected to the laptop
authorities. or the personal computer using a serial to universal serial bus
(USB) convertor.
 If the candidate is found to be fraudulent, then stringent
actions are taken. Here, while implementing, the fingerprints have to be
enrolled. However, in the actual examination scenario,
 If the alert is found to be false (due to system errors, fingerprint data can be extracted from the database of Aadhaar
which may occur sparingly), after a thorough verification Card.
by the higher officials, the examination can be resumed.
Using the Aadhaar Card details, the identity of the person
can be verified. Hence, Aadhaar card forms an integral part of
this method of impersonation detection, as it reduces the
resources required during the enrolment phase as well.
Fig. 7 shows the process of finger verification, where a finger
is placed on the optical fingerprint scanner. Fig. 8 shows the
software that recognizes the fingerprint as a valid one and
returns its identification (ID) as the result. When the fingerprint
recognition is successful, a system is allotted to the student and
Fig. 4. Instructions for the examination. the student has to take that system.

Fig. 5. Alert Message displayed on the screen.

Fig. 7. Fingerprint verification system.

Fig. 8. Fingerprint verification system.

Fig. 8 .Result of fingerprint identification.

Fig. 6. Alert message sent.

C. Face Recognition is stored as the dataset against which the face of the candidate is
Face recognition is the step after face detection. Here too, a compared during the course of the examination.
face has to be detected and distinguished from the background.
After face detection, the face has to be recognized to check if it
is the actual student.
1) Viola-Jones Face Detection
The algorithm that is used here for the purpose of face
detection is Viola-Jones Face Detection algorithm. It mainly
consists of four stages: Haar features detection, finding the Fig. 11. Dataset of the Face image captured when fingerprint is
verified.
integral image, adaptive boosting (AdaBoost) to reduce the
computational complexity and Cascade classifying the image In Fig. 12, the registered candidate is shown taking the
using combinations of weak and strong classifiers so as to examination. When an unknown face is found to be in view for
eliminate the chances of false rejections as much as possible. An more than a minute, then the examination is paused, as shown in
example of face detection using this algorithm is shown in Fig. Fig. 13. Once the examination pauses, the alert message is sent
9. to the higher authorities.

Fig. 12. Face recognition system recognizes the trained face and
Fig. 9. Face Detection (an example). allows the examination to continue.

2) Local Binary Pattern Histogram (LBPH) Recognition

Of methods like the Eigen Faces (based on Principal
Component Analysis), Fisher Faces (based on Linear
Discriminant Analysis) and LBPH, LBPH is chosen as it is more
robust and focuses on important features of the face, rather than
taking the entire face as a single entity [7,8].
LBPH is less affected by the pose variations and the light
intensity variations. During an examination, pose variations and
light intensity variations are inevitable, hence LBPH would be
an appropriate face recognition algorithm. An example of face
Fig. 13. Face recognition system does not recognize the untrained
recognition using LBPH is shown in Fig. 10. face and pauses the examination. The screen shows an alert
message.

IV. RESULTS AND DISCUSSIONS

An automated method of detecting impersonation in online
examinations has been developed, which requires minimal
human intervention and does not depend on the trustworthiness
of the invigilators present in the examination hall. After the
fingerprint verification is done, the student is allowed to take
his/her seat, where a screen asking for the username and
password is displayed. After logging in, at regular intervals, the
face detection is done, and if any extra face or any mismatch in
the face is found multiple times, then the examination is paused
Fig. 10. Face Recognition (an example)
and an alert SMS is sent to the examination authority in-charge
of that particular center.
As shown in Fig. 11, a set of images of the face of the Due to the rapid developments in the field of biometric
candidate is captured when fingerprint verification is done. This recognition, especially that of face recognition, wherein, three-
dimensional identification is being evolved, this method of [3] S. G. Anuradha, B. Kavya, S. Akshatha, K. Jyothi, G. Ashalatha,
identification would help eliminate impersonation. “Automated face detection and recognition for detecting impersonation
of candidate in examination system”, International Journal of Scientific &
In the case of unavailability of proper fingerprint (like Engineering Research, Vol. 7, Issue 3, ISSN 2229-5518 (March 2016)
accident victims, or certain skin disorders), iris recognition can [4] N. A. Karim, Z. Shukur, “Using preferences as user identification in the
be done. Aadhar Card has information about the iris and the online examination”, International Journal on Advanced Science
Engineering Information Technology, Vol. 6, No. 6, ISSN 2088-5334
fingerprint, hence the details can be verified using the same. (2016)
In the near future, the online examination would replace the [5] K. Patrick, Dr. S. O. McOyowo, Dr. H. O. Okoyo, “Addressing
pen-and-paper methods, and hence many robust methods to impersonation threats in online assessment environment using temporal
information and systems interactions”, Merit Research Journal of
eliminate impersonation could further be developed using Education and Review, Vol.3, pp. 215-220, ISSN: 2350-2282 (June 2015)
biometric and physiological behavior patterns of the individuals. [6] J. W. Gathuri, A. Luvanda, S. Matende, S. Kamundi, “Impersonation
challenges associated with e-assessment of university students”, Journal
REFERENCES of Information Engineering and Applications, Vol.4, No.7, ISSN 2225-
0506 (2014)
[1] Hindustan Times (January 2011), “Online tests gain traction in India” [7] M. Pietikainen (2010), “Local binary patterns” [Online]. Available:
[Online]. Available: http://www.livemint.com/Politics/ 4n7yjM4f t1aa http://www.scholarpedia.org/article/Local_Binary_Patterns (Accessed:
S6fqhl rB4I/Online-tests-gain-traction-in-India.html (Accessed: January February 2017)
2017)
[8] OpenCV 2.4.13.2 documentation, “Face recognition with OpenCV”
[2] Aakash, “The war of offline vs online exams show rising popularity of [Online]. Available: http://docs.opencv.org/2.4/modules/contrib/doc/
new trends in education” [Online]. Available: http:/ facerec/facerec_tutorial.html (Accessed: February 2017)
/aakashinstitute.aakash.ac.in/blog/the-war-of-offline-vs-online-exams-
show-rising-popularity-of-new-trends-in-education/ (Accessed: January
2017)