Hashing and Watermarking in Wireless Data

It is very easy to the tamper with digital data, like audio, images and video without leaving any clues. Under these circumstances, integrity verification has become an important issue in the digital world.

Traditional solutions for media verification are built around watermarking. Watermarking, is a media authentication/protection technique that embeds invisible (or inaudible) information into the media. For content authentication, the embedded watermark can be extracted and used for verification purposes.

Vishal Monga's dissertation work proposes an alternate solution to the media verification problem via "secure perceptual hashes". Unlike watermarking, hashing does not embed anything in the media but instead extracts, in a secure manner, media features that are representative of its content. Further, perceptual hashes are distinguished from conventional cryptographic hashes in that they only respond to perceptually meaningful changes, hence being able to perform verification of media content in a manner consistent with human perception. Traditional cryptographic hashes on the other hand, change dramatically even as a single bit in the input message is changed. Vishal's research combines signal processing and cryptographic approaches to build robust as well as secure image hash functions.

Vishal is currently spending the summer at Microsoft Research in Redmond, WA working on multimedia anti-piracy problems. He plans to continue his research in media security following his graduation in August 2005.