Research Of Digital Image Watermarking Algorithm Based On Zigzag And Discrete Wavelet Transform

With the computer network technology and digital technology worldwidepromotion and use,images,videos and other multimedia digital works become verypopular on the network. Though the digital works spread very fast on the network,they are not completely safe. Because the network multimedia digital works can beeasily copied and spread, pirates can make full use of these characteristics to do someillegal things，which threat to the author’s copyright seriously.Therefore,it is necessaryfor us to take some technical means to protect digital works. The digital watermarkingtechnology have come into being. As an important branch of information hidingtechnology research,it should own good security, transparency and robustness.This paper is based on discrete wavelet transform for digital image watermarkingalgorithm research.First of all, it elaborate the digital image watermarking researchbackground, significance and the situation at home and abroad. Then, it introduces theconcept and basic principle of digital watermarking. After analysising the commonscrambling technology, the wavelet transform is introduced.On that basis, three different algorithms based on wavelet transform arerecommended. The three algorithms are first to scramble the watermark in order todecrease the spatial correlation by Arnold Transform，then the watermark is one-leveldiscrete wavelet transformed.The first algorithm,the original image is two-levelwavelet transformed and embed the new watermark into the even rows and evencolumns of the original image’s low and intermediate frequency sub-band.Afterreversing wavelet transform,the watermarked image is extracted. The secondalgorithm is similar to the first algorithm, the difference between them is we choose toembed the new watermark into the odd rows and odd columns of the original image’slow and intermediate frequency sub-band. The third algorithm，the original image istwo-level wavelet transformed,we choose to scramble its low and intermediatefrequency sub-band N times by Zig-zag scrambling.Then we embed the newwatermark into the even rows and even columns of the new original image’s low andintermediate frequency sub-band.After that we should choose to scramble thewatermarked image N times by Zig-zag. Lastly, by reverseing wavelet transform,the watermarked image is extracted. Result of some common attack tests can prove thetheoretical analysis and achieve the desired results.