Fragile watermarking techniques for image authentication

Abstract

Techniques to establish the authenticity and integrity of digital images are becoming increasly essential over the Intermet. The authentication algorithm should distinguish malicious manipulations from the original watermarked image. Fragile watermarking is recomended for a very sensible image authentication scheme, able to localize the slightest changes occured to the marked image. The fragile watermarking shows the limits of its applicability, when secure transacting of digital content is required. In this case, the authentication algorithm should be able to distinguish incidental integrity maintaining distortions such as lossy compression from malicious manipulations. This has motivated further researches into semi-fragile watermarking. Two novel watermarking algorithms are proposed in this thesis, both robust to self-authenticating. The first algorithm is a fragile spatial parity-checking watermarking method that employs a public-key scheme for still image authentication and integrity verification. The use of the parity-checking scheme and the public-key encryption enables the localization of tampered areas and the authentication of the watermarked image respectevely. The second algorithm is a semi-fragile holographic watermarking system. The previous algorithm is a really powerfull system to be a fragile technique but still not usefull for secure transacting over the Internet. The holographic technique enables good perfomances against distortions and is still able to guarantee the authenticity and the integrity of digital images. The authenticity is achieved by a public-key encryption of the mark and the integrity is achieved by the holography introduced in this technique. In addition, the thesis provides theoretical analysis for the perfomance and the feasibility of both schemes.We also present experimental results to verify the theoretical observations and the comparison results for both algorithms to popular techniques.