IJSART - Volume 7 Issue 7 – JULY 2021 ISSN [ONLINE]: 2395-1052

Fraud Detection In Banking Using Key Agreement

And Face Authentication
Miss. Poonam Khape1, Miss. Rukmini Thite2, Miss. Komal Sale3, Guide:Prof. Waghmode. B.B4, Prof.Pandhare.S.D5.
1, 2, 3,4, 5 Dept
of Computer Science and Engineering
1, 2, 3, 4, 5 Punyashlok
Ahilyadevi Holkar Solapur University, Solapur of college
Sahakar Maharshi Shankarrao Mohite Patil Institute of Technology And Research Shankarnagr, Akluj

Abstract- Nowadays, online payments are concerned about current finish, we tend to propose associate degree
security. This is mainly due to hacking of OTPs or PIN codes appearance-flow-based face formalization convolutional
by the hackers. This paper proposes a method for credit card neural network(A3F-CNN). Specifically, A3F-CNN learns to
system used in transaction system which will integrate with determine the dense correspondence between the non-frontal
the face detection. The problem faced by credit card users is and frontal faces. Once the correspondence is constructed,
lot of privacy issues. This may commonly occur when users frontal faces area unit sized by expressly “moving” pixels
give their credit card numbers to unfamiliar individuals or from the non-front alone. In this way, the artificial frontal
when cards are lost. Our solution proposes a technique by faces will preserve fine facial textures. To improve the
which the image clicked by using webcam during the convergence of coaching, associate degree appearance-flow
registration process image captured detect the face and with guided learning strategy is planned. Additionally, generative
feature extraction face stored into the database. When user adversarial network loss is applied to attain a lot of photo
login the shopping portal that time system check previous realistic face, and a face mirroring technique is introduced to
transaction through IP address or mac address if they found handle the self-occlusion drawback. in depth experiments area
the address then process further other wise system discard the unit conducted on face synthesis and create invariant atomic
transaction. At the time of payment made by the user on E- number 87. Results show that our technique will synthesize a
commerce portal will be compared to the features from the lot of photo realistic faces than the existing ways in each the
dataset of the respective user and also process the two step controlled and uncontrolled delighting environments.
verification process. Face Features extracted from the images Moreover, we tend to reach aawfully competitive FR
stored in administrator main database. Our goal is to performance on the Multi-PIE, LFW and IJB-A databases.
compare the similarity of faces embedded in the credit card
and respective user. II. LITERATURE REVIEW

Keywords- Webcam, Transaction, Verification, Two Step 1.Sorournejad, Zahra Zojaji, Reza Ebrahimi Atani, Amir
Verification, Mac address, OTP, Face recognize, Image Hassan Monadjemi, “A Survey of Credit Card Fraud
Processing, etc. Detection Techniques: Data and Technique Oriented
Perspective”.
I. INTRODUCTION
Credit card plays a very important rule in today’s
In this Paper Facial create variation is one among the economy. It becomes an unavoidable part of household,
main factors making face recognition (FR) a difficult task. business and global activities. Although using credit cards
One standard solution is to convert non-frontal faces to frontal provides enormous benefits when used carefully and
ones on that FR is performed. Rotating faces causes facial responsibly, significant credit and financial damages may be
picture element price changes. Therefore, existing CNN-based caused by fraudulent activities. Many techniques have been
ways learn to size frontal faces in color house. However, this proposed to confront the growth in credit card fraud. However,
learning drawback in an exceedingly color house is very non- all of these techniques have the same goal of avoiding the
linear, inflicting the artificial frontal faces to lose fine facial credit card fraud; each one has its own drawbacks, advantages
textures. during this paper, we tend to take the view that the and characteristics. In this paper, after investigating
non frontal-frontal picture element changes area unit basically difficulties of credit card fraud detection, we seek to review
caused by geometric transformations (rotation, translation, and the state of the art in credit card fraud detection techniques,
so on) in house. Therefore, we tend to aim to find out the non datasets and evaluation criteria.The advantages and
frontal-frontal facial conversion within the spatial domain disadvantages of fraud detection methods are enumerated and
instead of the color domain to ease the training task. To the compared. Furthermore, a classification of mentioned

Page | 291 www.ijsart.com

IJSART - Volume 7 Issue 7 – JULY 2021 ISSN [ONLINE]: 2395-1052

techniques into two main fraud detection approaches, namely, each group on the base of all behavioral patterns. Finally, we
misuses (supervised) and anomaly detection (unsupervised) is use the classifier set to detect fraud online and if a new
presented. Again, a classification of techniques is proposed transaction is fraudulent, a feedback mechanism is taken in the
based on capability to process the numerical and categorical detection process in order to solve the problem of concept
datasets. Different datasets used in literature are then drift. The results of our experiments show that our approach is
described and grouped into real and synthesized data and the better than others.
effective and common attributes are extracted for further
usage. Moreover, evaluation employed criterion’s in literature 4. S. G. Fashoto, O. Owolabi, O. Adeleye, and J. Wandera,
are collected and discussed. Consequently, open issues for ”Hybrid methods for credit card fraud detection using K-
credit card fraud detection are explained as guidelines for new means clustering with hidden Markov model and multi-
researchers. layer perception algorithm”.

2. V. Bhusari1, S. Patil, “Application of Hidden Markov The use of credit cards is fast becoming the most
Model in Credit Card Fraud Detection”. efficient and stress free way of purchasing goods and services;
as it can be used both physically and online. Hence, it has
In modern retail market environment, electronic become imperative that find a solution to the problem of credit
commerce has rapidly gained a lot of attention and also card information security and also a method to detect
provides instantaneous transactions. In electronic commerce, fraudulent credit card transactions. Over the years, a number
credit card has become the most important means of payment of Data Mining techniques have been applied in the area of
due to fast development in information technology around the credit card fraud detection. The focus of this paper is to model
world. As the usage of credit card increases in the last decade, a fraud detection system that would attempt to maximally
rate of fraudulent practices is also increasing every year. detect credit card fraud by generating clusters and analyzing
Existing fraud detection system may not be so much capable the clusters generated by the dataset for anomalies. The major
to reduce fraud transaction rate. Improvement in fraud objective of this study is to compare the performance of two
detection practices has become essential to maintain existence hybrid approaches in terms of the detection accuracy.
of payment system. In this paper, we show how Hidden
Markov Model (HMM) is used to detect credit card fraud 5. V´eronique Van Vlasselaera, Cristi´an Bravob, Olivier
transaction with low false alarm. An HMM based system is Caelenc, Tina Eliassi-Radd, Le man Akoglue, Monique
initially studied spending profile of the card holder and Snoecka, Bart Baesensa, “APATE: A Novel Approachfor
followed by checking an incoming transaction against Automated Credit Card Transaction Fraud Detection
spending behavior of the card holder, if it is not accepted by using Network-Extensions”.
our proposed HMM with sufficient probability, then it would
be a fraudulent transaction. In last decade, the ease of online payment has opened
up many new opportunities for e-commerce, lowering the
3.Changjun Jiang, Jiahui Song, Guanjun Liu, “Credit geographical boundaries for retail. While e-commerce is still
Card Fraud Detection: A Novel Approach Using gaining popularity, it is also the playground of fraudsters who
Aggregation Strategy and Feedback Mechanism”. try to misuse the transparency of online purchases and the
transfer of credit card records. This paper proposes APATE, a
With the rapid development of electronic commerce, novel approach to detect fraudulent credit card transactions
the number of transactions by credit cards are increasing conducted in online stores. Our approach combines (1)
rapidly. As online shopping becomes the most popular intrinsic features derived from the characteristics of incoming
transaction mode, cases of trans-action fraud are also transactions and the customer spending history using the
increasing. In this paper, we propose a novel fraud detection fundamentals of RFM (Recency Frequency - Monetary); and
method that composes of four stages. To enrich a card holder’s (2) network-based features by exploiting the network of credit
behavioral patterns, we first utilize the cardholders’ historical card holders and merchants and deriving a time-dependent
transaction data to divide all cardholders into different groups suspiciousness score for each network object. Our results
such that the transaction behaviors of the members in the same show that both intrinsic and network-based features are two
group are similar. We thus propose a window-sliding strategy strongly intertwined sides of the same picture. The
to aggregate the transactions in each group. Next, we extract a combination of these two types of features leads to the best
collection of specific behavioral patterns for each cardholder performing models which reach AUC-scores higher than
based on the aggregated transactions and the card holder’s 0.98..
historical transactions. Then we train a set of classifiers for

Page | 292 www.ijsart.com

IJSART - Volume 7 Issue 7 – JULY 2021 ISSN [ONLINE]: 2395-1052

III. RESULT ANALYSIS

This system contains two modules:

1. Admin:

This module first create user account with the help of

user information and send account details to the user
registered mobile number by text message.

2. User:

Here user login his account do any type of transaction.

If user register successfully .

IV. ADVANTAGES

1. It is more secure than previous work .

2. Faster than manual paper work process.
3. Two step security like key agreement and face
authentication.

Page | 293 www.ijsart.com

IJSART - Volume 7 Issue 7 – JULY 2021 ISSN [ONLINE]: 2395-1052

4. No need of extra hardware to capture face for system far better. Many variants of the project can be
identification purpose. developed and utilized for home security and personal or
5. Reliable and user friendly. organizational benefits.

V. APPLICATIONS VIII. CONCLUSION

Proposed systems offer several Applications: – Our proposed project has been designed for the
purpose of reducing the credit card frauds that may occur
• Banking Sector. during online payment transaction. There is no need of
• Healthcare Sector. specialized hardware for installing this system. It just need a
• Organizational Sector. computer and a camera for construction. The system is reliable
and efficient mode of transaction process. Purpose of Our
VI. ALGORITHMS proposed project on credit card authentication using face
recognition is reducing credit card frauds that may occur
1. Viola Jones: during an online payment process. It should be flexible so that
people can easily use it without any hesitation. The web cam
1. Set the minimum window size, and sliding step plays an important role in the system. It is reliable and user
corresponding to that size. friendly. Using this technique solve the issues by integrating
2. For the chosen window size, slide the window face Recognition ,this system still lacks the ability to identify
vertically and horizontally with the same step. At people with similar face. This problem is overcome by using
each step, a set of N face recognition filters is OTP. The comparison of the real time image with the images
applied. If one filter gives a positive answer, the face stored in the database should be reliable.This system should
is detected in the current widow. not be made wait long time for a user.
3. If the size of the window is the maximum size stop
the procedure. Otherwise increase the size of the REFERENCES
window and corresponding sliding step to the next
chosen size and go to the step 2. [1] Zhihong Zhang , Xu Chen, Beizhan Wang, Guosheng Hu
, Wang- meng Zuo , and Edwin R. Hancock, ”Face
2. LBP: Frontalization Using an Appearance-Flow-Based
Convolutional Neural Network”, IEEE TRANS-
1. Parameters: the LBPH uses 4 parameters: Radius ACTIONS ON IMAGE PROCESSING, VOL. 28, NO. 5,
Neighbors Grid X Grid Y. MAY 2019.
2. Training the Algorithm. [2] Yudong Guo, juy-ong zhang, Jianfei Cai, Boyi Jiang,
3. Applying the LBP operation. Jianmin Zheng, ”CNN-Based RealTime Dense Face
4. Extracting the Histograms. Reconstruction with Inverse-Rendered Photo- Realistic
5. Performing the face recognition. Face Images”, IEEE Transactions on Pattern Analysis and
Machine Intelligence Volume: 41 , Issue: 6 ,June 1 2019.
VII. FUTURE WORK [3] Kirti Dang, Shanu Sharma, ”Review and Comparison of
Face Detection Algorithms”, 978-1-5090-3519-
• Since we have proposed a modular approach we can 9/17/31.00c 2017 IEEE978-1-5090- 3519-9/17/31.00c
improve different modules until we reach an acceptable 2017 IEEE.
identification and authentication rate. The comparison of [4] Shweta Jamkavale, Ashwini Kute, Rupali Pawar, Komal
the input image and the image stored in database should Jamkavale, Prashant Jawalkar, ”Secure Transaction By
be fast and reliable enough. Using Wireless Password with Shufing Keypad”, Volume
• The same project can be utilized for several security 4 Issue X, October 2016 IC Value: 13.98 ISSN: 2321-
purpose applications where authentication is needed to 9653.
access the privileges of the respective system. It can be [5] Ranjeet Singh1, Mandeep Kaur, ”Face Recognition and
used in recognizing and detecting guilty parties involving Detection using Viola-Jones and Cross Correlation
in unauthorized business. Face recognition and face Method”, International Journal of Science and Research
detection algorithm can be improved with respect to the (IJSR) ISSN (Online): 2319-7064 Index Copernicus
utilization of resources so that the project can recognize Value (2013): 6.14 — Impact Factor(2013):4.438.
more number of faces at a time which can make the

Page | 294 www.ijsart.com

IJSART - Volume 7 Issue 7 – JULY 2021 ISSN [ONLINE]: 2395-1052

Guide Name: prof. Waghmode B. B

Student name: Ms. Thite R. A .
College name : Smsmpitr, Akluj
College name : Smsmpitr, Akluj.
Mail id : babasaheb27@ gmail.com
Mail id : [email protected]
Qualification: ME computer engg. Phd CSE*
Dept name : comp science and engg.

Student Name: Ms. Sale K. R.

Collage name : smsmpitr ,Akluj .
Mail id : [email protected]
Dept name : comp science and engg.

Student name : Ms. Khape P. P.

College name : Smsmpitr, Akluj.
Mail id : [email protected]
Dept name : comp science and engg.

Page | 295 www.ijsart.com