SPREAD SPECTRUM WATERMARKING METHOD FOR IMAGE DATABASES

摘要

By means of the Internet, the image distribution may be easily and quickly done. However, a lot of scepticism may be found in this respect: an unlimited number of unauthorised copies can be obtained with quite banal equipment. This paper aims at presenting a new watermarking method which is able to afford ownership and track down copy maker. Our main contribution consists in the way in which we embed a very low power SS-CDMA (Spread Spectrum - Code Division Multiple Access) signal in the most significant DCT coefficients. In contrast to other SS-based approaches mentioned in the literature, our method complies with the StirMark robustness requirements. Note that we embed a 64 bits message, thus making it possible to identify a copy maker or, at least, a group of copy makers, depending on the targeted application. Although the most of the experimental results reported in the present paper correspond to grey level still image databases, the method may be extended to any sequence of images; for instance, we obtained very good results when watermarking some grey level video sequences.