Research Of Dual-Watermarking Algoirthm Based On Spatial And Frequency Domain

In the network，the digital multi-media data can be very easily copied, faked and broadcast, theproblems of protecting the data copyrights come out.Watermarking technology, which can solve theproblem effectively, attracts widespread attention and becomes an important research topic. But, atpresent, most watermark algorithm embed the single watermark to protect the copyrights, but theprotection and test ability of this sort of algorithms are limited, which can not achieve the multi-protection and multi-verification..The main work of this paper is using the Contourlet Transform and PCNN to decompose andpreprocess the image, then, embedding the dual watermarkings in image to achieve the multipurposeprotection and multi-verification. Specifically, the paper sets the static image as the research object andproposes two kinds of algorithm based on frequency and spatial domain:(1) The first dual-watermarking algorithm is based on Contourlet Transform and Image Fusion. Thealgorithm uses the image fusion theory to embed the visible watermark in the local carrier image andembeds the invisible watermark into the Contourlet frequency domain. The innovations are as follows: a)introducing a coordinates into visible watermarking image in order to correct the carrier image beingattacked by geometric attacks; b) using the odd-even quantization to embed the invisible watermark intothe block that is complicated texture and low degree of sensitivity,so that it can improve the invisibilityand robustness in the meanwhile. This algorithm can enlarge the embedding strength greatly and keep theimage visual quality well, otherwise, it can achieve the copyright multi-verification because of the visualwatermark.(2) The second dual-watermarking algorithm is based on Contourlet Transform and PCNN, it iszero-watermark concept. In Contourlet frequency domain, constructing zero-watermarking by comparingthe coeffs value to the average value; in the spatial domain, comparing the points value obtained by PCNNand their average value to construct the zero-watermarking. The innovations are as follows: a)using thePCNN to extract feature points, and construct the points template to correct the carrier image suffered bygeometric attacks; b)using the zero-watermark concept, constructing the zero-watermarking by the logiccomparison of relevant variable. This algorithm, low complexity, can settle the contradiction betweenrobustness and non-visibility basically, moreover, it can resist various kinds of attacks to achieve multi-protection.