14 DAY TRIAL // An electrical engineer at the New Jersey Institute of Technology (NIJT) has managed to crack a code that will allow investigators around the world to know for sure if and where a picture has been modified. Finding out if a photo has been changed will be very simple, NIJT professor of electrical and computer engineering Yun-Qing Shi, the lead developer of the new software, says. All it will take will be a computer, the photo and the software, and the identification will be automatically done, so the person analyzing the photograph will basically have to do nothing, except wait for the results.
“Using our program, we can usually inspect a photograph on a computer screen and know that someone has changed it. We still cannot say, nor can anyone else, where in the media the image has been changed. But we will get there,” Shi explains. Over the past few years, the expert has worked relentlessly to invent new and better ways of detecting changes to images. With four patents already granted to him, and about two dozen others pending, Shi has gained international recognition and lectures on the subject at universities around the world.
“In our digital age, digital media has been massively produced, easily manipulated, and swiftly transmitted to almost anywhere in the world at any time. While the great convenience has been appreciated, information assurance has become an urgent and critical issue faced by the digital world,” the information assurance and digital data forensics expert adds. Over the past couple of years, companies from the multimedia-security research sector have become increasingly interested in the kind of work Shi is conducting, so it's possible to see his technology applied very soon.
Users who post pictures on the Internet should be very careful to leave their mark on them, he stresses. While methods such as crediting the photographer or adding watermarks also work, the expert suggests that people could use electronic signatures for their media, a piece of coding that is generated automatically. “The signature is generated electronically and then one of many available methods can be employed to hide it,” Shi shares.
“Thanks to our new patents, if a user hasn't embedded identifying information, we will still be able to detect a forged image,” he tells, referring to situations where a picture is deemed to be “airbrushed,” but its origins are unclear. A Los Angeles Times photo of the Iraqi War, taken in 2003, and a BBC News image of the Israeli air strike against Beirut, in 2006, are those that have caused the most stir-up, as they have been proven to have been doctored. Shi's work is focused entirely on making sure that such situations do not occur in the future. If they do, we will at least be able to identify the modifications.