Zero-Watermark Algorithm Based On Lifting Wavelet And Iris Image Preprocessing

Recently, various kinds of digital multimedia data have been widely spreaded via Internet. As the commercial price of multimedia data is increased and the resulting multimedia data can be easily duplicated via network, the importance of copyright protection for multimedia data becomes more crucial. Digital watermark technique is embedding some information (the watermark) into multimedia data, but not affecting the commercial value of the original content. It should be invisible to human eyes. According to the information hidden in the multimedia data, owner of the copyright can affirm the ownership of the digital product or identify the integrality of content. The watermark can be the author's serial number, corporation identification, significant text, etc.Digital watermarking has been studied in the spatial domain and frequency domain. The spatial-domain watermarking algorithm is generally fast and simple, but it doesn't guarantee that the watermarking would be robust against noises and JPEG compression. The frequency domain watermarking algorithm is embedding watermark in frequency domain by using the character that strong signal can cover up weak one, and its robustness is much better than the spatial-domain watermarking algorithm. At present, the common frequency domain watermarking algorithm primarily adopts wavelet transform or DCT.At present, almost digital watermarking methods embed watermark by changing the original information. Such as modify the least significant bits (LSB), the corresponding coefficients of transform (DWT, DCT or FFT).But modify excessive coefficients, the visual effect is bad, modify few coefficients, the robustness is not strong. Zero-watermark method (Wen Quan [1]) does not change the original image, and construct watermark from features of image. So it solves the conflict between robustness and invisibility. In this paper, we select some coefficients of lifting wavelet transform as the features of the public image. It uses classical Cryptology methods to embed watermark. The grey value of the image is integer, so it is unavoidable to introduce error due to transformations from real number to integer. Therefore, we propose a watermarking method based on integer wavelet transform, all calculations are from integer to integer, so it avoids the error above. Further more, it accelerates the computing speed.In the first part of this paper, a zero-watermark method of embedding the watermark in digital image based on the lifting wavelet transform is proposed, the adopted wavelet is 5-3 lifting wavelet. It uses methods of classical Cryptology. And it does a series of experiments. Moreover, it has been proved have strong robustness by results of experiments.In age of electronic information we need a more convenient and effective identity verification technology. Biological recognition technology makes use of the inherent physiological or behavioral characteristics of human beings to identify individuals. Due to the unique, stable and live physiological properties of the iris and non-invasive to users, most stable and reliable for identification in practical applications of iris-based system, iris recognition is receiving extensive attention and becoming an active topic in biological recognition technology.The image below is an iris image in the database, the black and circular area in the centre is pupil, the circle bigger than pupil is the boundary of sclera, the area between pupil and sclera is iris.In the second part, this paper describes methods about iris image preprocessing including localization and normalization, main job is the localization of pupil and sclera. It brings a localization method based on difference of radial direction. The circle make C maximum is sclera. ( a , b ) is centre of the circle, r is radius,α<β,△θ=π/(80), I is iris image after processing. directions, the infection of eyelid and eyelash is minimum, so calculates difference in the radial directions.This paper does the correspond experiment with CASIA Iris Database (Version 1.0), and gets good results.