Digital Watermarking for Image Ownership Protection on Internet

Abstract

This research proposes a robust watermarking strategy for copyright protection that embeds multiple copyright information in a selected set of DCT coefficients. During watermark extraction, we introduce the majority-voting scheme that can render more reliable watermark detection compared to most conventional image watermarking approaches. The proposed method extracts the embedded watermark from the watermarked image without using the original image, so it is feasible for implementation on the Internet. To further accommodate the characteristics of different images, the proposed algorithm uses an adaptive watermark strength factor to strike a good balance between robustness and perceptual invisibility. The proposed algorithm solves the traditional watermarking problem of security degradation due to publication of the watermarking algorithm. Experimental results demonstrate that the watermark is robust to most signal processing operations and geometric distortions. It is found that detection capability, imperceptibility, and robustness of the watermarking technique can be guaranteed.