An algorithm for optical character recognition of the embossed digits on credit cards

Guides:

  • Mr. Raghavendra Hosabettu, Senior Consultant, Chief Technical Office, Wipro Technologies, Bangalore
  • Mr. Anil Kumar Lenka, Senior Consultant, Chief Technical Office, Wipro Technologies, Bangalore

The main motivation of this project was to make online payment easy and hassle free for the consumers, by developing something that assists them in online transactions, while still maintaining their privacy. Present day payment techniques involve swipe or manually entering the card number. We plan to change this approach by allowing the user to take an image of the card using the mobile application (app), thus eliminating the need for any hardware as well as the hassle of manually entering the digits.

The end goal of this project was to develop a elegant solution for online transactions for the end users, without the need of any additional hardware. This app was intended to help the end users during online payments, wherein, instead of entering the 15 or 16 digits of their card number, they can simply take an image of the card and the application will, in turn, extract the digits from the image acquired.

The proposed algorithm was able to detect the digits with an accuracy of 83%, with experiments carried out on cards issued by different corporations such as VISA, MasterCard, Maestro, American Express and RuPay.

Tools used: C++, Python, OpenCV, MATLAB, RoboRealm