Research On Anti-distortion Holographic Watermarking Algorithm And Its Application Based On SIFT Feature Points And DWT

With the rapid development of computer network,the diversity of image transmission on the Internet and its extensive application make piracy and copyright disputes increasingly serious social problems.Digital watermarking technology has become an effective means of intellectual property protection and digital multimedia anti-counterfeiting.At present,more and more watermarking algorithms have been proposed and developed at a very fast speed.In the process of application,people have higher and higher requirements for the robustness and anti attack ability of the watermark.In recent years,anti geometric distortion watermarking algorithms have gradually emerged,but few of these watermarking algorithms can resist nonlinear geometric attack,which is still in the initial stage.In view of this situation,this paper proposes anti-distortion holographic watermarking algorithm based on SIFT feature points and DWT.Firstly,the original watermark information is transformed into holographic watermark by Fresnel diffraction transformation.Then SIFT feature points algorithm is used to extract the feature points of the carrier image,and the distribution characteristics of feature points are analyzed.According to the difference of embedding region,two kinds of embedding algorithms are proposed: one is to select the 16 × 16 area around the feature point as the center,decompose the area by wavelet transform,block the holographic watermark and embed it in the low-frequency component in turn.The other is to use the distribution structure of the feature points to realize the triangle block of the carrier image,and then embed the holographic watermark into triangular blocks according to the appropriate embedding strength to realize the embedding of the hologram watermark in the carrier image.The embedding and extraction of the two algorithms are realized through programming.The results show that there is almost no difference between the carrier image and the carrier image with the watermark,and the extracted watermark image is also very clear.Because of the introduction of the holographic watermark and the use of the two transform domain watermarking algorithms,the watermark has the ability to resist attacks such as cropping and noise,has strong robustness and good results.For watermarked carrier images with nonlinear geometric deformation,a watermark correction algorithm based on block fuzzy approximation is proposed,which approximates the nonlinear geometric attack of the whole image to the linear geometric attack of small regions,and realizes the correction of the whole image through local correction.Firstly,the feature points of geometrically deformed watermarked carrier image and the original carrier image are extracted and matched respectively.Triangular blocks are divided according to the corresponding relation of matched feature point pairs,and the transformation matrix of each triangle area is found by the corresponding relationship of the triangle coordinates.Then the gray-scale pixels are reconstructed by image interpolation method to obtain all pixel values of the corrected triangular region and the corrected distorted image.In this paper,nonlinear geometric attacks are tested on watermarked carrier images,including spherical attacks,stretching attacks,cosine attacks,etc.Experimental results show that the proposed correction method can extract watermark information more clearly,and it can also extract the watermark for a certain degree of non-linear geometric distortion(attack),which effectively improves the robustness of the watermark,and the method has better performance of anti geometric distortion.