Today, mass-scale electronic content distribution systems embed forensic tracking watermarks primarily at the distribution server. For limiting the bandwidth usage and server complexity and enhancing scalability, it is preferable to embed the watermark at the client. Embedding in these untrusted clients requires secure embedding methods that do not leak unmarked content or the watermarking secrets. In this work, we propose a secure watermark embedding scheme based on lookup tables for spread-spectrum watermarks, which are robust to noise and can be detected without comparison to the original content. We also develop fast detection mechanisms that make the watermark detection feasible for tracking systems with a large number of clients. Our fast detection algorithm improves detection speed of existing methods by six orders of magnitude in a typical system with millions of clients.

In a forensic watermarking architecture, a buyer-seller protocol protects the watermark secrets from the buyer and prevents false infringement accusations by the seller. Existing protocols encrypt the watermark and the content with a homomorphic public-key cipher and perform embedding under encryption. When used for multimedia data, these protocols create a large computation and bandwidth overhead. In this correspondence, we show that the same functionality can be achieved efficiently using recently proposed secure watermark embedding algorithms.

In an electronic content distribution system, it is preferable to embed forensic tracking watermarks at the client-side to limit bandwidth usage and server complexity. Embedding in these untrusted clients, however, requires secure embedding methods that do not leak unmarked contents or the watermarking secrets. In this work, we propose a look-up-table (LUT) based cipher, similar to Andersen's Chameleon cipher, for securely embedding spread-spectrum watermarks, which are noise robust and detectable without the original content. We also develop fast detection mechanisms that make the watermark detection feasible for tracking systems with large number of clients. Our fast detection algorithm improves detection speed six orders of magnitude in a typical system.

In this paper we present a reversible watermarking technique for digital images. In our context, reversibility refers to the ability to restore the original image by the watermark detector. The technique is based on a transformation function that introduces 'gaps' in the image histogram. The 'gaps' are used to encode the watermark. Several experiments revealed that a relatively high embedding rate may be achieved. For a set of test images we obtained embedding rates between 0.06 and 0.6 bits per pixel at PSNR levels of 45-50.