[Student] [Professor's Performance] Professor Woo Simon's lab undergraduate student, Moon Hak-Joon won the Excellence Award for th
- 소프트웨어융합대학
- Hit834
- 2023-07-10
Moon Hak-joon (1st author, In Bachelor of Software) and Park Eun-joo (In Ph.D. in Artificial Intelligence) of the Data-Based Convergence Security Laboratory (DASH Lab: https://dash-lab.github.io Professor Usaiman) won the Excellence Award at the Korea Information Protection Association Summer Conference.
This study proposes a method of determining the authenticity of an identification card using a neural network to confirm the user's identity with high accuracy in non-face-to-face situations.
Summary: Mobile identity authentication systems are widely used mainly for e-commerce and digital banking. In order to use the service non-face-to-face, identification cards such as resident registration cards and driver's licenses are photographed in the process of authenticating the user's identity. However, since it is not possible to confirm that the actual ID card is being photographed with the user's camera, it is necessary to determine the authenticity of the photographed ID card. In this paper, deep learning techniques were used to determine whether the user's remotely provided ID image was real or manipulated in a non-digital area (an image printed in high definition or an image printed on a monitor after shooting). In addition to RGB images as input from the model, we experimented using discrete Fourier transform and feature extraction techniques. When the authenticity of the ID image was determined using the learned model, a classification accuracy of up to 96.6% was achieved.
Paper Name: A Study on the Manipulated Identification Method Using Artificial Neural Networks
Author: Moon Hak-joon, Park Eun-joo, Kim Jung-ho, Yoon Kwan-sik*, Seo-yeon*, Woo Simon Seong-il
*: Samsung SDS