Abstract
Watermarking is a new technique for copyright protection. The robustness of watermarking is desirable in most practical application. In this paper, we propose a statistic-based watermarking scheme for color images in discrete wavelet domain. Unlike the traditional methods, the watermark is embedded into the statistic of the wavelet coefficients instead of an individual coefficient. We also present a channel adaptive watermarking (CAW) and a strength control function (SCF) to balance the robustness and the invisibility of the watermark. By experiments our algorithm shows a high performance on surviving multiple attacks from Stirmark Benchmark 4.0. The algorithm extends to blind watermarking easily, although the original watermark is needed in this paper.