Bilkent University

Department of Computer Engineering

S E M I N A R

Robust Signal Hashing for Images and Audio

Kıvanç Mihcak

Microsoft

Seattle, WA, USA

In this talk, we consider the multimedia hashing problem for images and audio: Given multimedia content as an input, the "signal hash function" should produce a short bit string (termed hash value) such that the hash values are similar (if not the same) for "perceptually similar" content and significantly different for "perceptually different" content. Furthermore, for anti-piracy applications, the hash functions should be secure. We begin the talk with the definition of the problem of robust signal hashing and some application scenarios. Then, we present three hash algorithms (two for images and one for audio) together with some experimental results. The first algorithm uses Singular Value Decomposition (SVD) on randomly chosen regions of images. The second algorithm collects randomized linear statistics of audio signals in the time-frequency domain. The third algorithm produces hash values of images via iterative filtering techniques. We conclude with the future work and some open problems.

DATE: March 14, 2005, Monday @ 13:40

PLACE: EA 409