Research On Digital Watermarking Algorithm In Contourlet Transform Based On Human Visual System

With the increasingly widespread application of digital products, digitalinformation security issues become issues of concern, of which the copyright protectionof digital products the most important and urgent. As an effective means of ensuring thesecurity of digital watermarking, digital watermarking become the hot spot in industryand academia. However, the security of digital watermarking remains to be furtherimproved, and how to maximize the watermark embedding strength in order to enhancethe robustness of digital watermarking, under the premise of ensuring invisibility, is alsostill unresolved problem.Considering the two shortcomings of Arnold scrambling, on its basis a newimage-scrambling algorithm is designed in this paper, which is that using a non-definedtransformation matrix to scramble the watermark image by Arnold transform, then thealgorithm second encrypts the watermark image, with the pseudo-random sequencefrom fractional calculus of sinusoidal signal. According to the basic principles andcharacteristics of Contourlet transform, combining with the Human Visual System, adigital watermarking algorithm in Contourlet transform based on Human Visual Systemis designed. This paper has improved the wavelet-based model of Human Visual Systemto adapt to the Contourlet transform, and the visual masking model in Contourlettransform is achieved with the improvement of the three aspects of visual maskingmodel cover factor, brightness and texture cover factor. And then watermarkinformation is embedded by multiplicative principle, according to the maximumembedding strength calculated by Just-Noticeable-Difference.The algorithm designed in this paper second encrypts a watermark image on thebasis of Arnold scrambling, which further improved the security and robustness whilemaintaining a high scrambling effect by Arnold transform, through eliminating thegenerality and periodicity of Arnold transformation. And it taking full advantage of theContourlet transform, combining with human visual features, chooses the appropriatedirectional image to embed watermark, which ensure that the watermark information isinvisible in the premise, and watermark embedded in the maximum intensity.Simulation results show that the designed algorithm has strong robustness undercommon image processing and geometric transformations attacks.The next work will continue to focus on how to further improve the capability toresist a variety of attacks, and to further increase the amount of watermark information.