For Doctored Photos, A New Flavor Of Digital Truth Serum
Distinguishing between authentic digital photos and doctored images is critical for law enforcement, the military, and newspapers and magazines, to name just a few affected areas. Dartmouth College computer science Professor Hany Farid has developed algorithms that can tell the difference between genuine and altered digital images: They can, for example, detect fine details that are signs of spliced or rotated images.
Tests on several hundred doctored pictures showed that the method was practically infallible, provided the picture was of sufficient quality. Farid says that files created in the highly popular JPEG format were harder to grade accurately.
Meanwhile, State University of New York at Binghamton Professor Jessica Fridrich has embarked on an Air Force-funded project to design a camera that takes two pictures simultaneously; one of the photos would be of the photographer’s iris, and this image would be compressed, encrypted, and buried within the larger photo taken through the camera lens, along with the time and place the photo was snapped.
The photo’s authenticity would be called into question if the encoded data is missing or tampered with: “It establishes a connection between the person who took the image, the camera used, and the digital image itself,” Fridrich explains.
Farid does not think digital watermarks are a secure enough solution, and acknowledges that preventing the doctoring of digital images may ultimately be a futile goal. He states, “At the end of the day, the person doing the tampering has the easier job. And they’ll win.”
Abstracted by the National Law Enforcement and Corrections Technology Center(NLECTC) from the New York Times (07/22/04) P. E5; Shachtman, Noah .