A digital watermark is a short sequence of information containing an owner identity or copyright information embedded in a way that is difficult to erase. We present a new oblivious digital watermarking technique for copyright protection of still images. The technique embeds the watermark in a subset of low to mid frequency coefficients. A key is used to randomly select a group of coefficients from that subset for watermark embedding. The original phases of the selected coefficients are removed and the new phases are set in accordance with the embedded watermark. Since the coefficients are selected at random, the powers of the low magnitude coefficients are increased to enhance their immunity against image attacks. To cope with small geometric attacks, a replica of the watermark is embedded by dividing the image into sub-blocks and taking the DCT of these blocks. The watermark is embedded in the DC component of some of these blocks selected in an adaptive way using quantization techniques. A major advantage of this technique is its complete suppression of the noise due to the host image. The robustness of the technique to a number of standard image processing attacks is demonstrated using the criteria of the latest Stirmark benchmark test.

Date Published: 1 August 2001
PDF: 9 pages
Proc. SPIE 4314, Security and Watermarking of Multimedia Contents III, (1 August 2001); doi: 10.1117/12.435386

Published in SPIE Proceedings Vol. 4314:
Security and Watermarking of Multimedia Contents III
Ping Wah Wong; Edward J. Delp III, Editor(s)