Photo Authenication

In March of this year, the Duke University community was shocked when members of the Duke Lacrosse team were accused of raping a 27-year old student attending North Carolina State University. Apparently, the woman was originally hired as a stripper to perform at a party.

Of interest here is the fact that defense attorneys are attempting to use time-stamped digital photos to construct an alibi for at least one of the defendants. Certainly, during the course of the upcoming trial the credibility of such a time stamp will be an issue. The time stamp embedded in digital photographs is only as good as the accuracy of the internal clock and the original setting of the time. Moreover, it does not take much computer savvy to modify the time in a digital image file. Indeed, in most cameras the time is stored in plain text format. The credibility of the time stamps will in part be a function of how soon the camera was seized by authorities and the consistency of the time stamps on other photographs with some sort of fiducial references.

This last week, the blog site Little Green Footballs alerted the blog community of a conspicuously altered photograph published by the Reuters News Service. Apparently, a free lance photographer, Adnan Hajj, had clumsily duplicated buildings and smoke on an aerial shot of Beirut making it appear that the damage cause by Israeli bombing in Lebanon was more extensive. Later it was discovered that another one of Hajj’s photographs had been modified. An Israeli jet was identified as firing rockets, when it had instead launched a flare. The flare and its trail were been duplicated in the photo to suggest more aggressive action by the Israeli Air Force than captured on the image. Reuters fired the photographer and pulled his photographs. More recent analysis suggests that Hajj’s “Photoshop” sins are perhaps outweighed by unmodified, but staged photographs.

In view of these developments, the question arises as to whether digital photographs could be self-authenticating. Is it possible to design a camera that would mark a photograph in a way that would make any tampering difficult to hide? One would expect that journalistic organizations would require such equipment. Given inexpensive and wide spread availability of photo-editing software, it would lend credibility to their photographs. Moreover, if the time stamps could be automatically syncing to time standards, then the reconstruction of timelines would more authoritative. If such a technology became ubiquitous, then common commercial cameras could aid in the prosecution or vindication of legal suspects.

One method suggested by Kodak is to use a public and private key encryption scheme to encrypt a message summary in each photo. The photograph could be read by any conventional software and the message summary read with the public key. The file could be copied and modified for any purpose, but any modification of the original would make the encrypted message summary inconsistent with the image, indicating that the file had been changed. The fudging of photographs by the likes of Mr. Hajj would be made far more difficult if possible at all.

However, the time stamp included in the message summary is only as good as the time setting of the camera. One could imagine cameras with no human time-setting function, but whose clocks were periodically updated with GPS satellite signals. Of course, it would be possible to pull batteries and zero out the time and take a photograph before there was an opportunity to sync the time. Fortunately, such deliberate circumvention attempts would be conspicuous and invalidate self-authentication, making such an effort less valuable for alibis and photojournalism.

For photo-journalism, we will still be largely dependent upon the honesty of photojournalists, because there is no way for a camera to provide authentication that a photograph was not staged. For this we must rely on the scrutiny of editors, who at least in the case of Reuters and these photographs appears unprofessionally credulous. Hajj was correctly fired. We hope that in the near future, Reuters will punish the professional photo-editors that could not spot clearly manipulated images that took the blogosphere only a short time to debunk. Nor did Reuters detect clearly staged photographs. If the blogs had not performed this service, it is likely that these phony photos would still be circulating at the Reuters news service.

Leave a Reply

You must be logged in to post a comment.