Protocols for Watermark Verification

deter piracy of multimedia content, the owner typically reveals the watermark in the process of establishing piracy. Once revealed, a watermark can be removed. We eliminate this limitation by using cryptographic protocols to demonstrate the presence of a watermark without revealing it. C onsider an application where multi-media content is electronically distributed over a network. To discourage unauthorized duplication and distribution, the content owner can embed a unique watermark (or a fingerprint) in each distributed copy of the data. If the owner later finds unauthorized copies of the data, then the origin of the copy and the identity of the erring buyer could be determined by retrieving the unique watermark corresponding to each buyer. These schemes are sometimes called copy deterrence watermarking schemes or digital fingerprinting schemes. We'll focus our attention on digital image watermarking, although the same problems exist for other multimedia data, such as video or audio. A watermark is a signal added to the digital image that can later be extracted or detected to make an assertion about the image. 1 Two types of watermarks exist: visible and invisible. Visible watermarks typically contain conspicuously visible messages or company logos indicating the ownership of the image. Invisible watermarks, on the other hand, are unobtrusive modifications to the image and the invisibly watermarked image visually appears similar to the original. Users can determine the existence of an invisible watermark only by using an appropriate watermark extraction or detection algorithm. Companies generally prefer invisible watermarks as their unobtrusiveness makes them more desirable. We can also classify watermarking techniques as fragile and robust. Any image processing procedure will corrupt a fragile watermark, whereas a robust watermark can resist common image manipulation procedures (such as rotation, reflection , scaling, cropping, smoothing, contrast or brightness adjustment, or lossy compression). Clearly, a watermark used for the purpose of copy deterrence must be robust. Yet another classification of watermarking techniques is into oblivious and nonoblivious schemes. A nonoblivious scheme requires an original or reference image in the watermark detection procedure. On the other hand, an oblivious scheme doesn't require the use of an original or reference image. Thus, oblivious schemes are attractive for many applications. In copy deterrence watermarking schemes, the watermarks used are generally invisible, robust, and oblivious. Recall that we deter copying by inserting a unique watermark into each copy of the image sold (which we can use to trace unauthorized copies to …