Tuesday, January 11, 2011

Final Year Project Sharing: Image Forensics Toolkit

Project Title: Image Forensics Toolkit
Categories: Computer Forensics, Digital Forensics, Image Forensics
Students: Yeo Quan Yang, Tan Boon Siong, Kang Xian Yang, Ng Jun Xuan, Lee Jun Ting
Organization: I2R
Supervisor: Samson Yeow (SP), Dr Ng and Dr Gao (I2R)

Synopsis: The purpose of the project is to explore the use of recaptured data and to develop counter-measures to deal with facial recognition attacks with photos

Aim: It aims to create data that can help in detecting image of images and an infrastructure to research and facilitate on capturing recaptured images. The data which we use consists of images of scenery and facial images and they are recaptured using many kinds of devices. Some of them would be the Acer M-900 phone, HP iPAQ hw 6960, Nokia N95 and the iPhone 3GS. As for the infrastructure, it will be a website which hosts data samples of recaptured images that we have collected for users to use as research data.

In this project, the group found out that illumination and camera angles are very crucial factors in good data samples. The quality and clarity of the recapturing device is also very important in recapturing images. As for the website, the group decided to let people download the dataset so that researchers may use it if they are keen on this direction of research too because image recapture datasets are not widely available on the internet now.

Commenting on his learning experience, student Yeo Quan Yang said, "Through this project, I learn that we should be clear of what we are doing and also the scope of the project so that we would understand what we are doing and avoid doing things that are out of the scope. Teamwork is also very important, as we need everyone to complete their task on time so that we can move on to the next task."

"Taking recaptured image is not an easy task as well. I have to consider the wind and brightness of the photo. Location is an important factor. It has to be neither too bright nor too dark. At the start, my hand would be shaking a lot when taking the photos. Now, I can take very quickly and without much error." said Tan Bong Siong.

Another student, Kang Xian Yang said, "If I have not taken this project, Image Forensics Toolkit, as my final year project, I will most likely have not thought of using pictures to bypass facial recognition authentication systems. I feel this is an ingenious way of hacking. This shows that hacking is not just limited to the programming level. And there are other ways of hacking."

No comments:

Post a Comment