Dartmouth Researcher Investigates Digital Secrecy

Working with Dartmouth's Institute for Security Technology Studies, Farid focuses on scouring digital images for traces of hidden messages. His research, important to the legal community, aims to expose suspiciously altered images. People engaged in political and corporate espionage or illegal pornography, for example, have learned to send secretly embedded messages in their vacation or holiday photos through e-mail.

"The courts are terribly unprepared to handle the new breed of digital criminals that has emerged along with the rapid increase in low-cost and sophisticated digital technology. As the criminals get smarter, so must we", said Farid.

Steganography, the art of concealing and sending messages, has been around as long as people have had secret information to relay. This practice has come a long way since the days of letters with invisible inks carried by midnight messengers or encrypted Morse code delivered over secret radio frequencies. Computers and the World Wide Web provide a new twist on this covert activity. Today's digital cameras produce high-quality images, and the Internet easily and inexpensively carries enormous volumes of information worldwide.

"To expose modern steganography, the first step is to discover a way to detect if there is a hidden message embedded within an image. We can't yet decode the message, but we can flag an image as suspicious," said Farid.

A digital image is a collection of pixels, and each pixel contains numbers that correspond to a color or brightness value. With high-resolution images, it's easy to hide a message by slightly altering the numbers, called LSB, or least significant bit, manipulations. The changes to the image are imperceptible to the human eye, and the secret information goes undetected.

Employing the mathematics of wavelets, a popular method for compressing digital data, and progressive image coding, Farid's research began by characterizing statistical image properties that are consistent across most images. After embedding secret messages into the images, Farid showed that these same statistics are fundamentally changed. This process led to the creation of a computer program that can determine the likelihood that a secret message has been hidden within an image. So far, Farid's program is 90 percent accurate.

Farid's work also has applications in other fields. For instance, the technology could be beneficial in digital forensics, the science of authenticating digital material for use as evidence in court. The research might also assist art historians, collectors or gallery owners trying to detect art forgeries. Farid thinks that paintings can be scrutinized mathematically and compared to authentic paintings, to determine if the same artist created them.