A Cryptography based Face Authentication

System for Secured Communication

2022 IEEE Fourth International Conference on Advances in Electronics, Computers and Communications (ICAECC) | 978-1-6654-0239-2/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICAECC54045.2022.9716676

Ajeethra BA Gautham Prasath SV Arun Balaji R

School of Computer Science and School of Computer Science and School of Computer Science and
Engineering Engineering Engineering
Vellore Institute of Technology, Vellore Institute of Technology, Vellore Institute of Technology,
Vellore, India Vellore, India Vellore, India
[email protected] [email protected] [email protected]

Kakelli Anil Kumar

Associate Professor, School of
Computer Science and Engineering
Vellore Institute of Technology,
Vellore, India
[email protected]

Abstract— Digital communication and networking had are two technologies used to enhance security levels.
become an integral part of our everyday life. Technological Cryptography aims to provide secured communication by
advancements in Digital networking must also include security allowing only the sender and the intended recipient to view the
and confidentiality paradigms. Several previous works on content of the message encryption and decryption. There are
communication systems comprised the problem of storage, many cryptographic algorithms are existing at present with
sharing, and complexity of keys. On considering the mentioned their specifications and advantages. On considering various
problems of existing works, this paper proposes secure factors and previous research works, the Rivest, Shamir,
communication using cryptography and face recognition Adleman (RSA) algorithm is used in this model to secure the
techniques with cloud computing. The proposed system pertains
message due to time factors and also the level of a competency
to a protected communication process, where messages are
entitled only after the verification of the authorized sender and
[1][2].
receiver using Linear Binary Pattern Histogram (LBPH) face Face authentication gives a lending hand in securing data
recognition, and Rivest, Shamir, Adleman (RSA) cryptographic by recognizing the authorized person by using various
technique with the cloud management system. The system techniques. There are various Face Recognition algorithms
generates RSA key pair, which is exported as a Privacy- prevailing and the Linear Binary Pattern Histogram (LBPH)
Enhanced Mail (PEM) file and stored in a remote server algorithm is used in this model as it served to be the most
through a Secure Shell (SSH) tunnel.The proposed system has
suitable when compared to other algorithms [3]. Along with
found that using 50 samples for face authentication is most
these, we have used Secret Shell (SSH), a cryptographic
efficient and accurate with limited time. Existing works have
focused to increase security by adding layers of encryption which
protocol to operate network service securely [4]. This is being
in turn increased the complexity to handle keys and decryption used in this model to obtain a secure and authenticated
processes. This proposed methodology on following a biometric pathway to the cloud.
authentication system, stretches itself with an extra efficient
layer of security without increasing the complexity of the system.
II. RELATED WORK
A. Image Recognition
Keywords— LBPH face recognition, RSA cryptographic
Kriti Jain et.al have discussed the comparison between
algorithm, Secure Communication, SSH tunnel
commonly used algorithms for face recognition. LBPH was
I. INTRODUCTION found to be most accurate with less noise interference when
compared to other algorithms tested. Concluded by proving its
In today’s fast-evolving world, digital networking has accuracy even on dynamic light conditions during the
become the most integral and essential part of our life. We recognition [3]. Anjali T et.al have depicted a face recognition
human entities have given ourselves so much into digital system using Linear Binary Pattern Histogram (LBPH) and
communication and networking in such a way that it’s almost Haar Cascade techniques. Tkinter module is being used to
impossible to lead our everyday life without it. While create Graphical User Interface (GUI) applications in this
wondering and appreciating numerous advantages of digital system and resulted in an accuracy rate of 60% by processing
communication, we tend to forget its disadvantages on the a 3x3 matrix [5].
flipside. Digital communication where millions of users fetch
their data is prone to various attacks. It is a distressing factthat Fayaz Ali et.al have achieved an improvised face
the enhancement of technology brings along with itself the recognition model at a Low-resolution minimum of 35 px, to
attackers and hence are more prone to various attacks. Secured capture face at various angles and even during human motion
communication is the most important requirement of every using LBPH. Dataset (LR 500) is being designed for training
domain over the years of digitalization. Hence improvements and classification and consists of three main processes as Face
in digitalization must incorporate more security measures to Representation, Face extraction, and classification. The input
sustain its advantages. Cryptography and Face Authentication of face gestures is achieved by the Haar Cascade classifier.
Face extraction is achieved by using LBPH histogram and is

978-1-6654-0239-2/22/$31.00
Authorized licensed use©2022
limitedIEEE
to: Zhejiang University. Downloaded on February 22,2025 at 00:53:23 UTC from IEEE Xplore. Restrictions apply.
being compared with the dataset (LR500) to classify and R.K. Tiwari et.al addressed a technique in which they encrypt
analyze the image whether or not it belongs to the picture from data using the RSA algorithm to provide protection such that
the database [6]. S.S. Giri and V.S. Karwande have reviewed only the concerned user has access to it. User data would be
deep learning techniques and the execution of automatic encrypted before being transferred to the cloud. When a user
attendance systems using various other techniques. The requests data, the Cloud provider verifies the user's identity
images are captured and positioned using the Haar cascade and delivers the data [15]. Rajan.S. Jamgekar et.al discusses
method and feature extraction is done later in the process. The an improved version of the RSA algorithm for achieving
face detection process is also done by the Haar cascade security during file transmission. Secure RSA is used during
method and the detected part is then cropped for further file transfer which removes few loopholes of RSA and thus
classification. The Recognition part of the face is a node in a prevents from attackers misusing the data [16]. ESI Harba
3x3 matrix using the LBPH Algorithm [7]. focuses on providing a high-security system by combining
three types of encryption methods such as RSA, AES, and
B. Cryptographic Encryption Using RSA HMAC thus requiring low computation yet simple and fast
G. Geeta Sai Sruthi and M. Raghupathi have achieved the [17].
shortening of time in encryption and decryption of RSA by C. Client-Server Protocol
following three methods. Modular Multiplication and
Exponentiation Algorithms, RSA decryption, and RSA Key SSH gives an open protocol for obtaining a network
Management [8]. Abhishek Guru et al. focused more on a communication which costs lesser than VPN solutions. File
secured version of the RSA encryption algorithm by using transfer, Command shell, and data tunneling services for
more than four prime numbers thereby increasing the TCP/IP applications are provided by SSH. User
complexity of variables. Hence this paper proves that Authentication, Host Authentication, Data Encryption, and
improvised RSA encryption using four prime numbers is more Data integrity are four security benefits of SSH as described
secure and better than just using two prime numbers as in the in the paper [18]. SSH is a client server-based protocol that
traditional approach [9]. uses SSH key pairs (public-private key pairs) for
authentication. It uses symmetric keys for encryption and
Michelle S Henriques and Nagaraj K. Vernekar have decryption where the keys are protected by asymmetric
achieved less encryption time using both asymmetric and approach and Hashing algorithms such as MAC and HMAC
symmetric encryption also it is more secure as random keys to validate the message for data integrity and is easily the best
generated do not have the relation between them [10]. Nasrin option for security [4]. On the result of research work done by
Khanezaei et.al gives the concealed way to deal with securing Irfaan Coonjah, Pierre Clarel Catherine, and K. M. S.
data in the distributed storage system by using a combination Soyjaudah, OpenSSH utilized the link better and improved its
of RSA and AES. It is used to secure sending and retrieving transfer times and speed compared to the Open VPN and is
of files in the cloud respectively. The limited-time efficient [19]. Maha A. Saya et.al has followed specific
consumption and having solved key distribution issues proved techniques to upgrade the security of Linux systems such as
the advantage of using Symmetric encryption Advanced authorized users and rate-limiting the connections. [20].
Encryption Standard (AES) in fetching files from the cloud.
The secret key of the AES algorithm generated by the cloud III. PROPOSED SYSTEM
earlier is being encrypted by the RSA algorithm which makes Providing security and confidentiality are the most
it impractical for attackers to attack [11]. challenging and critical needs of the digital communication
H. M. Leena et. al, proposed an efficient way of RSA system. This system serves the best for authenticated users to
implementation by using large selective prime numbers send and receive messages. As shown in Fig.1, our proposed
instead of the conventional way and also breaking the file into system includes cryptography, Face Authentication, and cloud
blocks and increasing the level of computation [1]. Iwasokun, techniques to enhance the security level of the communication
G. B et.al portrays the importance of financial data and how system. RSA algorithm is used for encryption and decryption
large it is for the cloud computing systems to process the data. [10] and the LBPH algorithm [21] for face recognition. The
Thus satisfactory mechanism must be set up to ensure the system can be categorized into three modules: Enrolment
safety of sensitive credit cards and other financial Data used module, Authentication module, and Message communication
in a Card-Not-Present (CNP) transaction [12]. Dr. Rajamohan module.
Parthasarathy et.al have researched securing the level of A. Enrollment Module
security obtained by RSA for the cloud data and protecting its
confidentiality [13]. The administrator collects the user’s information and
captures the user’s face to prepare the dataset. The data set
Pratiksha Gautam et.al have discussed that Electronic contains the User Name, User ID, and User’s face image. This
Health records and vulnerabilities of accessing data are the enrolment module is shown in Fig. 2 and undergoes the
key focus of this paper and the authors have used RSA and following steps.
three-layered architecture to protect the patient’s data and thus
achieving high-speed competency and security [14]. Ajit • Data Collection: Getting User ID and Username from
Kumar Santra et.al This paper uses a method of hybrid the user.
cryptography that attains double-level encryption and uses • Generation of RSA key Pair: RSA key pair is
RSA to store the key instead of the message. That specific key is generated using the User Name and User ID fetched by
used for the blowfish algorithm and at last Secure Hash the user in data.
Algorithm (SHA-2) algorithm is been used to provide a
message digest of encrypted data to keep data integrity [2].

Authorized licensed use limited to: Zhejiang University. Downloaded on February 22,2025 at 00:53:23 UTC from IEEE Xplore. Restrictions apply.
 Fig. 1. The proposed working model

• Storage of PEM file in cloud server: RSA key pair is • Dataset creation: Data set is created containing the
generated and exported as PEM file. PEM file is then User ID,User Name, and User’s face.
stored in a cloud server through an SSH tunnel which
is used to operate over a network securely. B. Authentication Module.
This module serves the purpose of authenticating the users
• Capture the User’s Face: Firstly, the user’s face is before the communication of messages and is shown in Fig. 3.
detected using the Haar cascade model in OpenCV The user’s face is detected and undergoes the face recognition
using python. Thereafter 50 samples of the user’s face process by authenticating with the dataset created during the
are captured and undergo a feature extraction process. enrolment module. Authentication is achieved by following
Each image is named in the format of “User ID. the two steps mentioned below.
Sample Number” and stored which is then used for
image recognition during the authentication process. • Recognition of User Face: The user’s face is detected
and the LBPH algorithm is used to recognize the user.
• Authentication process: Checks whether the user is
valid by accessing and comparing with the dataset
created in the enrolment module. If a user is enrolled,
then User Id and User Name will be displayed and
prompts whether to send or receive the message. If the
user is not enrolled, the process will be terminated.
C. Message Communication Module
Although the user is authenticated already using the face
recognition technique, communication of message among
users requires keys to send and receive the message which
makes the system more protected and secured. The complete
flowchart of this module along with the authentication module
is shown in Fig. 3. The user requires the keys to send and
receive the message which makes the system more protected
and secured. Sending and receiving messages are the two
processes in this module and are explained below.
• Send Message: The user enters the receiver’s id and the
message to be sent. The SSH connection is initiated
with the cloud server. Thereafter the user fetches the
receiver’s public key from the cloud server through
SSH to encrypt the message. The encrypted message is
sent and stored in a database that can be retrieved by
Fig. 2. The enrollment process the receiver.

Authorized licensed use limited to: Zhejiang University. Downloaded on February 22,2025 at 00:53:23 UTC from IEEE Xplore. Restrictions apply.
 Fig. 3. Authentication and message communication systems of the proposed system

• Receive Message: SSH tunnel is initiated to the cloud SSH authentication is then made to the cloud server. Once the
server by the receiver, to get the private key to decrypt cloud server verifies and authenticates the client machine, the
the message from the database. Once the private key is key file is stored on the cloud server. After such processes, 50
retrieved from the cloud server through SSH, the samples of the user’s face are captured by the system. The
message is decrypted and will be displayed to the captured image is trained and used for the authentication
receiver. process. Once the message is input, the system enquires for
SSH authentication with the cloud server to retrieve thepublic
IV. EXPERIMENT AND RESULTS key using the LBPH recognition system. Hence the User ID,
It is very important to secure the communication data, User Name, and Trained image are stored in the database as
both at transit and rest. Therefore, it is necessary to ensure the dataset for the future authentication process.
authentication and authorization steps are tamper-proof B. Phase 2
including when an unauthorized user tries to access the
system and its features. The main aim of this paper is to have The methodology here is the authentication and secure
an efficient crypto communication cloud system. We have communication. When the user wants to send or receive the
used Python to develop this system and hosted an internal message, the system first detects the user's face. The
cloud server to provide cloud functionalities. The system authentication process takes place by comparing the captured
includes two phases which are mentioned below. image with that of the stored image from the database and
recognize using the LBPH recognition algorithm. If the image
A. Phase 1 captured is not recognized, the system considers that the user
In the first phase, the user must be registered with the is not enrolled and notifies as an unregistered user and
system by using the User ID and Username as displayed in terminates the process. If the user is registered, the user types
Fig. 4. The system then creates a key file with the user id as ‘s’ to send the message. The system enquires for the receiver
the file name in the format of PEM and waits for SSH ID and the respective message as shown in Fig. 5. Once the
authentication. message is given, the system makes SSHauthentication with
the cloud server to retrieve the public keyof the receiver. The
public key is used to encrypt the message, the encrypted
message is stored in the database with the respective sender
id and receiver id. If the user wants to receive the message, ‘r’
is fed to the system. The system then initiates an SSH
authentication to the cloud server and retrieves the private key
of the user to decrypt the messages sent to the respective user.
The messages are read from the database which is sent to the
Fig. 4. User Registration respective user and are decryptedusing the private key.

Authorized licensed use limited to: Zhejiang University. Downloaded on February 22,2025 at 00:53:23 UTC from IEEE Xplore. Restrictions apply.
 Fig. 5. The message from the user

Fig. 6. The message received by the user

V. RESULT ANALYSIS AND DISCUSSION

The use of the LBPH algorithm to authenticate a user The message stored in the database is also encrypted using
provides efficiency and accuracy to the system. LBPH [21] is the receiver’s public key as observed in Fig. 10. Thus, the
chosen over other algorithms and its comparison is shown in system resists database attacks and ensures user’s privacy.
Fig. 7. As the number of samples increases as per Fig. 9,
accuracy increases, whereas the time required to authenticate
the face also increases as shown in Fig. 8. So, there is a trade-
off between accuracy and the time taken by the system to
authenticate the user [2]. On analyzing the trade-off between
time and accuracy, we found that a sample size of 50 proves
to be efficient and accurate enough for the system to
authenticate the user.
SSH protocol is used in this model to achieve secure
transmission of the key file to the cloud server. It ensures that
the system resists the Man-In-The-Middle attacks and
Sniffing Attacks. Security during data transit is verified by
capturing encrypted data packets to and from the server.
Fig. 8. No. of samples vs Time taken to authenticate a user

Fig. 7. The Face Recognition algorithms with their accuracies

Fig. 9. No of samples Vs the accuracy of the system

Authorized licensed use limited to: Zhejiang University. Downloaded on February 22,2025 at 00:53:23 UTC from IEEE Xplore. Restrictions apply.
 Fig. 10. Encrypted messages stored in database [Sender ID, Receiver ID, Encrypted Message]

VI. CONCLUSION AND FUTURE WORK

The main goals of our secure communication system are Devices in IOT’ IEEE 2017 International Conference on IOT and
to provide efficient authentication and secure the data both Application, 2017.
at rest and also in motion. The proposed system uses [11] Nasrin Khanezaei and Zurina Mohd Hanapi, 'A Framework Based
on RSA and AES Encryption Algorithms for Cloud Computing
biometric authentication which removes the barrier of using Services', 2014 IEEE Conference on Systems, Process and Control,
usernames and passwords thus enabling an efficient and 2014.
secure authentication process. Authentication using LBPH [12] Oluwole Charles Akinyokun, Sunday Julius Alawode and Taiwo
with a sample size of 50 has achieved a good trade-off Gabriel Omomule, ‘An RSA Algorithm for Securing Financial Data
between accuracy and time taken for the authentication on the Cloud’ Journal of Advances in Mathematics and Computer
process with RSA. Due to limited message size, the RSA Science, 2019 Pp.1-11.
algorithm has performed well as similar symmetric key [13] Dr. Rajamohan Parthasarathy P, Haw Wai Yee Mr. Seow Soon
Loong, Leelavathi Rajamanickam and Preethy Ayyappan,
algorithms. Efficient key management is achieved using 'Implementation of RSA Algorithm to Secure Data in Cloud
RSA in terms of limited key size, number, and storage Computing ', International Journal of Innovative Science,
space. Storing key files in a cloud server has provided secure Engineering & Technology, Vol. 6 Issue 4, 2019.
remote-authentication with secure storage for the keyfiles [14] Pratiksha Gautam, Mohd. Dilshad Ansari and Surender Kumar
where it is accessed only by the authenticated users of the Sharma, 'Enhanced Security for Electronic Health Care Information
system. Our proposed method has achieved secure Using Obfuscation and RSA Algorithm in Cloud Computing,
International Journal of Information Security and Privacy, 2019.
communication using biometric authentication with RSA
[15] Santosh Kumar Singh, Dr. P.K. Manjhi, Dr. R.K. Tiwari Data
and LBPH face recognition algorithm. This work can Security using RSA Algorithm in Cloud Computing ', International
further be improved by implementing the digital envelope Journal of Advanced Research in Computer and Communication
mechanism for the encryption and decryption to encrypt the Engineering, Vol. 5, Issue 8, 2016.
message using the symmetric key algorithm, and the [16] Rajan.S.Jamgekar, Geeta Shantanu Joshi, ‘File Encryption and
symmetric key is further encrypted by the RSA algorithm. Decryption Using Secure RSA’, International Journal of Emerging
Science and Engineering ISSN: 2319–6378, Volume-1, Issue-4,
REFERENCE 2013.
[1] Dr. D.I. George Amalarethinam, H. M. Leena,' Enhanced RSA [17] E. S. I. Harba ‘Secure Data Encryption Through a Combination of
Algorithm with varying Key Sizes for Data Security in Cloud ', IEEE AES, RSA and HMAC’, Engineering, Technology & Applied
2017 World Congress on Computing and Communication Science Research, Pp.1781- 1785 vol 7, Issue 4, 2017.
Technologies, 2017. [18] Nidhi Kandhil and Dr. Anil Kumar ‘A Study on Secure Shell (SSH)
[2] Divya Prathana Timothy, Ajit Kumar Santra, 'A hybrid cryptography Protocol’ International Journal Of Computer Science &
algorithm for cloud computing security, IEEE 2017 International Management Studies, Vol. 11, Issue 02, Aug 2011.
Conference on Microelectronic Devices, Circuits and Systems, [19] Irfaan Coonjah, Pierre Clarel Catherine and K. M. S. Soyjaudah,
2017. ‘Performance evaluation and analysis of layer 3 tunneling between
[3] Sudha Narang, Kriti Jain, Megha Saxena and Aashna Arora, openssh and openvpn in a wide area network environment’, IEEE
‘Comparison of Face Recognition Algorithms using Opencv for 2015 International Conference on Computing, Communication and
Attendance System’, International Journal of Scientific and Security, 2015.
Research Publications Vol.8 Issue.2, 2018. [20] Maha A.Saya, Naofal Mohamad hassin Azeez and rana sabah naser,
[4] Marko Aleksic, How Does SSH Work?, pnap KB ‘Improvement of cryptographic network protocol (SSH) for more
(phoenixnap.com). secure data communications’, International Journal of Recent
Innovation in Engineering and Research Publication, 2016.
[5] Bharath Tej Chinimilli, Vihas Reddy Kaipu, Anjali T, Akhil Kotturi,
Jathin Varma Mandapati, 'Face Recognition based Attendance [21] Manop Phankokkruad, Phichaya Jaturawat, ‘Effect of Variation
System using Haar Cascade and Local Binary Pattern Histogram Factors on the Processing Time of the Face Recognition Algorithms
Algorithm ', IEEE 2020 4th International Conference on Trends in in Video Sequence’, 2016 International Conference on Intelligent
Electronics and Informatics, 2020. Information Processing, Article No. 50, December, 2016.
[6] Afatb Ahmed, Jiandong Guo, Fayaz Ali, Farha Deeba and Awais
Ahmed, ‘ LBPH Based Improved Face Recognition At Low
Resolution’ International Conference on Artificial Intelligence and
Big Data, 2018.
[7] S.S.giri and V.S.Karwande 'Review Paper On Deep Learning
Techniques For Biometric Attendance System By Face Detection
Using Eye Ratina', Open Acces International Journal of Science and
Engineering, 2021.
[8] G. Geeta Sai Sruthi and M. Raghupathi ‘ A Secure Electronic
Messaging System in Client-Server Cryptography- RSA Algorithm’
International Journal of Engineering and Advanced Technology,
2019.
[9] Abhishek Guru and Asha Ambhaikar, ‘Development of RSA
Encryption Algorithm for Secure Communication’ International
Journal of Computer Sciences and Engineering, 2019.
[10] Michelle S Henriques and Nagaraj K. Vernekar, ‘Using Symmetric
and Asymmetric Cryptography to Secure Communication Between

Authorized licensed use limited to: Zhejiang University. Downloaded on February 22,2025 at 00:53:23 UTC from IEEE Xplore. Restrictions apply.