Research On Quantization-based Watermarking

A considerable amount of attention has been paid to a number of digital watermarkingmethods based on quantization lately, seeking to achieve in practice the capacity predictedby Costa for a channel with side information at the encoder. The first objective of thisdissertation is to study the performance yielded by quantization-based watermarking scheme.A further motivation is focused on non additive attacks, such as the gain attack plus noise,and the linear filtering attack. The main achievements are listed as follows:(1) We first consider the watermarking model as one of communications with sideinformation at the encoder. Potential performance improvement in watermarking systemwith side information at the encoder is analyzed. We also provide the performance analysisof Dither Modulation (DM) method, which is the special case of Quantization IndexModulation (QIM).(2) Statistical estimation techniques to combat amplitude scaling in combination withadditive noise within a Quantization-based watermarking context are proposed. First wederive the probability density function (PDF) of the watermarked and attacked data in theabsence of subtractive dither. Based on this model we propose a Fourier-based procedure forestimating amplitude scaling factors.(3) We also derive the PDF models for subtractive dither used at the encoder. Using anapproximation of the PDF model in the presence of subtractive dither, we derive a MaximumLikelihood (ML) procedure for estimating amplitude scaling factors.(4) We also study the problem of estimating two-band amplitude scaling on QIMwatermarking scheme. Quantization-based watermarking schemes are not robust againstLinear Time Invariant (LTI) filtering attacks. We concentrate on a multi-band amplitudescaling attack that modifies the spectrum of the signal using an analysis/synthesis filter bank.By extending the single channel estimation procedure, we propose a ML procedure forestimating two-band amplitude scaling attack. Experimental results on synthetic signalsdemonstrate the effectiveness of the scheme. (5) A quantization-based watermarking method, named Ratio Dither Modulation is studied.This scheme is based on using a gain-invariant adaptive quantization step-size at bothembedder and decoder. Experimental results prove the validity of keeping invariance againstamplitude scaling attack.(6) We also propose an extension of Ratio Dither Modulation to construct a scheme that isrobust to linear time-invariant filtering attacks, named Frequency Ratio Dither Modulationbasically works in the DFT domain.At last, some open issues in quantization-based watermarking as well as the future work arepresented.