Credit Card Transaction Based on Face Recognition Technology

This paper proposes a method for credit card transaction system which will make use of face recognition and face detection technology, using Haar Cascade and GLCM algorithm. The main problem faced by credit card users is attack to lot of privacy issues such as credit card. This generally happens when users give their credit card number to unknown people or when the card is lost. So, we are proposing a system that will reduce the risk of credit card frauds. The system we are proposing will match the image of user’s face with dataset of respective user. A database will be maintained for authentication purpose. If the image matches, that means user is genuine and he will be allowed to proceed otherwise, the user will be denied to do the transaction.


INTRODUCTION
In the current scenario, credit card and debit card are becoming the most common type of payment mode. All the credit card related task is managed by credit card processor.
Companies using credit card processing makes sure that transactions are processed correctly and on time. Many companies prefer online transaction because it benefits their business. Funds are transferred into account on time without putting much effort.
Since people are comfortable with cashless transactions, the demand of credit card is increasing rapidly. The main problem faced by the credit card users is to have a secure online transaction. Credit card fraud is a big challenge. The proposed solution will make use of face detection and face recognition technology for making credit card transaction system secured and much better.

RELATED WORK
One of the existing systems deals with securing payment process between the card issuer and the card reader terminal and it is thus ensures that card number is not known to any other entity, other than the two end points [1]. There is another system that suggests, a detection model must be available to capture the possible anomalous transaction [2]. There is a number of challenges like concept drift, class imbalance and verification latency in credit card fraud detection [3]. Machine learning can also be used for detecting credit card fraudulent transaction using a real world dataset [4] [5]. Deep learning presents a programming resolution to the matter of mastercard fraud detection that produces use of historic client knowledge in addition as real time group action details that area unit recorded at a similar time of group  [6]. There is another system that makes use of two random forests to train the behavior features of normal and abnormal transactions. The system compares the two forests which differs in classifier and analyze their performance on credit card transaction [7] [8]. One of the existing system makes use of Boat algorithm for detecting fraud transaction [9]. There is another system which analyze the periodic behavior of the time of a transaction using the Von Mises Distribution [10].

A. Haar Cascade
Haar Cascade, a machine learning object detection algorithm rule accustomed establish objects in a picture or video and supported the idea of options planned by Paul Viola and Archangel Jones. It's renowned for having the ability to observe (identify) faces and parts pictures. A Haar Cascade is largely a classifier that is employed to observe the thing that it's been trained for, from the supply. The Haar Cascade is by superimposing the positive image over a collection of negative pictures. The coaching is mostly done on a server and on numerous stages. Higher results are obtained by victimization top quality pictures and increasing the quantity of stages that the classifier is trained.
The algorithmic rule has four stages: Haar Feature Choice, making Integral pictures, Adaboost coaching, Cascading Classifiers. In general, 3 types of options are utilized in that the worth of a 2 rectangular options is that the distinction total of the pixels at intervals 2 rectangular regions. These regions have same form and size and are horizontally or vertically adjacent. Wherever as within the 3 rectangular options are computed by taking the total of 2 outside parallelograms then ablated with the total in an exceedingly center rectangle. Moreover, within the four rectangles feature computes the distinction between diagonal pairs of rectangles. First, an image of your face is captured from a photograph or video. Your face may seem alone or in an exceedingly crowd. Your image could show you wanting straight ahead or nearly in profile. After that, identity verification package reads the pure mathematics of your face. Key factors embody the space between your eyes and also the distance from forehead to chin. The package identifies facial landmark, one system identifies sixty eight of them, that are key to identifying your face. The result: your facial signature. Then your facial signature and a mathematical formula is compared to an information of farfarmed faces. And at last a determination is created. Your faceprint could match that of a picture in an exceedingly identity verification system information.

C. GLCM Algorithmic rule
It stands for Gray-level Co-occurrence matrix. The GLCM perform characterize the feel of a picture by calculative however usually pairs of pel with specific values and in an exceedingly specified spatial relationship occur in a picture, making a GLCM. A GLCM may be a matrix wherever the quantity of rows and columns is up to the quantity of grey levels, G, within the image.
The matrix components P (i, j | ∆x, ∆y) is that the frequency with that 2 pixels, separated by a pel distance (∆x, ∆y), occur at intervals a given neighborhood, one with intensity 'i' and also the different with intensity 'j'. The matrix components P (i, j | d, ө) contains the second order applied math likelihood values for changes between grey levels 'i' and 'j' at a selected displacement distance d and at a selected angle (ө). Employing a sizeable amount of intensity levels G implies storing a great deal of temporary information, i.e. a G × G matrix for every combination of (Δx , Δy) or (d, ө).

PROPOSED SYSTEM
The aim of the project is to implement a system, that uses face recognition and detection technique to authenticate the individual, in order to perform a successful and a secured transaction. This system will help in reducing credit card frauds. The aim is to make the transaction system fully automatic , that provides a reliable mode of online transaction process. In the proposed system, user gets authenticated by the system by matching the features of user image to the features stored in administrator. If the features matches, the transaction will be done successfully if not the user will be denied and transaction will not proceed.

MODULE DESCRIPTION
Module 1 (Admin Module)-Administrator register to the system using administrator login and password provided by the administrator is compared to the password stored in the system for the authentication purpose. This module is responsible for making changes in the system and get users registered to our system. In this, face image is cropped from the given image (input) the features are extracted and stored in the database using Gray Level Co-occurrence Matrix.

Fig. 4: Admin Module
Module 2 (User Module)-Here user authentication process is done. Only the authenticated user is allowed to proceed.
In this module, the password given by the user is compared to the password stored in the database. If it matches, user will be allowed proceed i.e. user will be allowed to do the transaction. If not he or she will be denied from doing the payment.    F. If the user is genuine, he will be allowed to do the transaction. This image shows the payment confirmation.