Application of composite invisible image watermarks to simplify detection of a distinct watermark from a large set

Abstract

Earlier, we presented a highly robust invisible watermarking technique for digitized images1 having a payload of one bit - indicating the presence or absence of the watermark. Other invisible watermarking techniques also possess this property. This family of techniques may be used to watermark a source image with distinct marks, perhaps to indicate the identity of the recipient, resulting in a set of many near-copies of the source image. Then, the problem of detecting a distinct watermark in an image from the set may imply attempting detection of all possible watermarks. In this paper we will present a technique using composite watermarks which reduces the number of attempts necessary for distinct watermark detection. If the number of images in the set is m to the power n, then the number of attempted detections is never more than m times n. Thus, for m = 10 and n = 3, a set of 1000 distinctly watermarked near-copies can be produced, but instead of 1000 attempted detection's to insure identification of a particular watermark, only thirty are required. The techniques used for constructing composite water-marks will be detailed and limitations of this approach will be discussed. Results of a successful detection of a distinct water-mark from a large set will be presented.