Watermark detection with zero-knowledge disclosure

Watermarking schemes embed information imperceptibly in digital objects and are proposed as primitives in various copyright protection applications, such as proofs of authorship, dispute resolving protocols or fingerprinting. In many applications, the presence of watermarks must be provable to any possibly dishonest party. Traditionally, watermark detection requires knowledge of sensitive information like the watermark or the embedding key. This is a major security risk, since this information is in most cases sufficient to remove the watermark and to defeat the goal of copyright protection. Zero-knowledge watermark detection is a promising approach to overcome security issues during the process of watermark detection: cryptographic techniques are used to prove that a watermark is detectable in certain data, without jeopardizing the watermark. This paper presents a formal definition of zero-knowledge watermark detection, discusses zero-knowledge watermark detection protocols and compares their properties.