Kit that detects fraud photos
DCE develops photo fraud detection kitindia Updated: Aug 18, 2006 03:31 IST
Two students of the Delhi College of Engineering (DCE) have developed a unique algorithm to counter problems of image morphing and digital forgery. They presented their project at the World Multi-conference on Systemics, Cybernetics and Informatics held at Orlando, Florida, in July this year. The duo, Sparsh and Himanshu, were the only two undergraduate researchers at the international conference.
"This presentation was a success and we are open to offers from the industry on this project. Since the work is completely manual, our future plan is to introduce automation," said DCE Principal Prof P.B. Sharma.
The algorithm is based on analysing the signature involuntarily inserted in a video upon a read-out from a digital camera.
This signature is then compared to similar signatures in various parts of the video to detect alien frames or areas that have been forged during video editing.
"The noise created by the forged video is compared to the noise from the original video. Then a differential is worked out to identify the forged patterns," Sharma said. This helps detect the alien frames inserted in the video.
This algorithm helps detecting two kinds of forgeries. One is of videos containing spurious frames, in which case the detection is carried out by comparing the frame noise to the average noise of the video. The other is where an area is forged in many frames and is detected by comparison of the noise present in a specific area to the average noise of the frame. A special case, in which an area is forged in only one of the frames among consecutive frames is also detected using this approach.
This algorithm was implemented on a forged video. A difference in noise pattern by about three per cent was obtained in a case where a video frame was forged using another video frame captured by the same camera. This three per cent difference was obtained between a forged frame and an original video frame.
First Published: Aug 18, 2006 03:31 IST