The Study Of Audio Watermarking Algorithm

With the development of electronic industrial technology，the spread and use of multimediainformation has become an indispensable part of daily life. However, the attendant problems areabout how to protect the copyright of multimedia information,how to transfer the secret informationand how to solve series of the related issues.According to this background,digital watermarkingtechnology emerged.Digital watermarking is a interdisciplinary subject which focus on the study ofcopyright protection of graph,audio,video and text.Our paper is about the study of audiowatermarking’s algorithm,the main work consist of the following aspects:(1) A novel method to encode the typical PN sequence was proposed,it is applied into a new echokernel for time-spread watermarking technique.The detecting method is to detect autocorrelation ofPseudo-random sequence and find the peak point,then judge the watermark bit.The typical methodbrings too much noise,and achieve not very good robustness.The encoded PN sequence reduce thenumbers of long‘1’and ‘0’and it’s autocorrelation has three peaks-a positive one and twonegatives.In this way,the new PN sequence result in smaller magnitude and less fluctuation in PSR(perceptually significant region).(2) The echo hiding watermark embedding approach is improved.According to psychoacousticmodel,a new self-adaptive embedding way was designed based on the concept ofDMOPM(Distortion Measurement On Psychoacoustic Model).It can reduce the noise effectivelyand choose the suitable coefficients,when embedding the watermarking bits.(3) Vector quantization (VQ) is widely used in speech recognition and photo compression.A zerowatermarking algorithm was proposed based on the Vector quantization.Select the discrete cosinetransform coefficients from the frames of audio,Principal component analysis (PCA)algorithm wasused to reduce the coefficient’s number.This method combine the discrete cosinetransform,Principal component analysis and BCH encoding into one.The experimental resultsdemonstrate that this algorithm achieve good robustness and detection reliability.(4) Propose a multipurpose watermarking approach.This approach design a multipurposewatermarking,which conclude a robustness watermark based on “zero watermarking” and asemi-fragile watermarking.The robustness watermark was embedded in the DCT coefficients withhigh energy,while the semi-fragile watermark was embedded in the coefficients with middleenergy.The semi-fragile watermark is based on the theory of index constrain vector quantization.The most important part in this approach is blocking the DCT coefficients andcalculate each block’s energy,then choose different part to be embedded.