Quantization Index Modulation Methods for Digital Watermarking and Information Embedding of Multimedia

TL;DR: This work presents a class of embedding methods called quantization index modulation (QIM) that achieve provably good rate-distortion-robustness performance, and introduces a form of postprocessing the authors refer to as distortion compensation that, when combined with QIM, allows capacity to be achieved.

Abstract: Copyright notification and enforcement, authentication, covert communication, and hybrid transmission applications such as digital audio broadcasting are examples of emerging multimedia applications for digital watermarking and information embedding methods, methods for embedding one signal (e.g., the digital watermark) within another "host" signal to form a third, "composite" signal. The embedding is designed to achieve efficient trade-offs among the three conflicting goals of maximizing information-embedding rate, minimizing distortion between the host signal and composite signal, and maximizing the robustness of the embedding. We present a class of embedding methods called quantization index modulation (QIM) that achieve provably good rate-distortion-robustness performance. These methods, and low-complexity realizations of them called dither modulation, are provably better than both previously proposed linear methods of spread spectrum and nonlinear methods of low-bit(s) modulation against square-error distortion-constrained intentional attacks. We also derive information-embedding capacities for the case of a colored Gaussian host signal and additive colored Gaussian noise attacks. These results imply an information embedding capacity of about 1/3 b/s of embedded digital rate for every Hertz of host signal bandwidth and every dB drop in received host signal quality. We show that QIM methods achieve performance within 1.6 dB of capacity, and we introduce a form of postprocessing we refer to as distortion compensation that, when combined with QIM, allows capacity to be achieved. In addition, we show that distortion-compensated QIM is an optimal embedding strategy against some important classes of intentional attacks as well. Finally, we report simulation results that demonstrate the performance of dither modulation realizations that can be implemented with only a few adders and scalar quantizers.