@phdthesis{FeiPhDThesis08,
title = {Analysis and Design of Watermark-Based Multimedia Authentication Systems},
author = {Chuhong Fei},
year = {2006},
date = {2006-01-01},
address = {Toronto, Canada},
school = {University of Toronto},
abstract = {Multimedia authentication techniques are required to protect the integrity and origin of digital multimedia data. In watermark-based multimedia authentication systems, a source-dependent authenticator is invisibly embedded within the protected data itself, in order to alert unacceptable distortions on the data and to authenticate the legitimate source. This thesis investigates the design and analysis of such systems. We formulate the authentication problem as error detection coding with robustness to certain types of legitimate channel noise. System performance requirements such as semi-fragility, security, key payload, embedding distortion, and algorithm complexity are all investigated in the unified framework to obtain best tradeoffs among them.In the analysis of multimedia authentication systems, we demonstrate how lattice codes can be used to address semi-fragile requirements and embedding distortion in a practical scheme. Security against possible malicious attacks is considered at the code level, and cryptographic techniques are incorporated to construct secure codes. We propose a novel security enhancement strategy based on nested lattice codes that results in secure and efficient authentication algorithms. Unconditional security is also investigated. We provide a formal methodology using composite hypothesis testing to balance robustness and fragility objectives with respect to legitimate and illegitimate distortions that is superior to existing approaches. We show that a quantization-based watermarking scheme is more appropriate for achieving semi-fragility.In system design, we propose two practical watermark-based authentication schemes for natural images: one for uncompressed images, the other for JPEG compressed ones. Our proposed algorithms are able to distinguish incidental distortions, such as JPEG compression and image enhancement filtering, from malicious tampering such as image object removal and substitution even in a small portion of the protected images. Our algorithms are secure against an active attacker, and are able to determine the degree of distortion and its location. In addition, our proposed joint compression and authentication scheme for compressed images is able to increase compression efficiency while keeping practical robustness to mild incidental signal processing changes. Effectiveness of the proposed algorithms is demonstrated by simulations and comparison with existing schemes.
In the analysis of multimedia authentication systems, we demonstrate how lattice codes can be used to address semi-fragile requirements and embedding distortion in a practical scheme. Security against possible malicious attacks is considered at the code level, and cryptographic techniques are incorporated to construct secure codes. We propose a novel security enhancement strategy based on nested lattice codes that results in secure and efficient authentication algorithms. Unconditional security is also investigated. We provide a formal methodology using composite hypothesis testing to balance robustness and fragility objectives with respect to legitimate and illegitimate distortions that is superior to existing approaches. We show that a quantization-based watermarking scheme is more appropriate for achieving semi-fragility.

In system design, we propose two practical watermark-based authentication schemes for natural images: one for uncompressed images, the other for JPEG compressed ones. Our proposed algorithms are able to distinguish incidental distortions, such as JPEG compression and image enhancement filtering, from malicious tampering such as image object removal and substitution even in a small portion of the protected images. Our algorithms are secure against an active attacker, and are able to determine the degree of distortion and its location. In addition, our proposed joint compression and authentication scheme for compressed images is able to increase compression efficiency while keeping practical robustness to mild incidental signal processing changes. Effectiveness of the proposed algorithms is demonstrated by simulations and comparison with existing
schemes.},
keywords = {drm, PhD thesis, thesis},
pubstate = {published},
tppubtype = {phdthesis}
}