Font Size: a A A

Research On Color Image Watermarking Algorithm Against Geometric Attacks

Posted on:2020-11-03Degree:MasterType:Thesis
Country:ChinaCandidate:D L MaFull Text:PDF
GTID:2428330596979678Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
The urgent need for digital media copyright protection has made watermark anti-geometric attack research a hot topic.In order to obtain a robust color image watermarking algorithm,which can resist geometric attacks well and have good watermark transparency,this paper studies the color image watermarking algorithm as follows:(1)A color image watermarking algorithm based on human visual characteristics(HVS)and Heisenberg decomposition(Hessenberg)is studied.The algorithm firstly performs 4*4 segmentation on the luminance component of the color image.Secondly,the information entropy and edge entropy of each image block are calculated by HVS characteristics and calculated the summed and sorted according to them;The block that the sum value is between 1/4 and 3/4 is used as the watermark embedding region;the Hessenberg decomposition is performed on the selected image block to obtain the upper Hessenberg matrix;finally,the maximum element value of the upper Hessenberg matrix is quantized to obtain the color image.The algorithm performs watermark embedding position selection by calculating the entropy value of the image block,and the selected position is evenly dispersed in each region of the image.Therefore,when the image after embedding.the watermark is cut,the watermark information is not easily destroyed,and the extracted watermark is complete;In addition,the algorithm uses an iterative embedding strategy to further improve its anti-shear attack.Experiments show that the algorithm can ensure that the watermark image has better visual effect and stronger robustness;especially for shear attack,the algorithm is more robust.(2)A color image watermarking algorithm based on anisotropic wavelet transform(Directionlet transform)and singular value decomposition(SVD)is studied.The algorithm first performs Canny edge extraction on the B blue component of the RGB carrier image,and then uses Hough transform to perform edge line detection to select a relatively continuous vertical or horizontal line;the selected line slope is used as the transformation direction of the Directionlet transform;then select the appropriate anisotropy ratio to perform Directionlet transform on the blue component image to obtain several sub-band images.The SVD transform is performed on the low-frequency sub-band with better robustness,and the singular matrix element value is modified to obtain a color watermark image.By using transformation characteristics of the Directionlet transform which along the arbitrary integer ratio slope direction,when the image is subjected to the rotation attack,the watermark embedding position can be effectively positioned according to the fixed angle between the fixed transform direction and the watermark embedding direction,thereby improving the anti-rotation of the watermark.The experiment proves that the anti-rotation performance of the algorithm is better under the condition that the watermark is not detectable.
Keywords/Search Tags:anti-geometric attack, HVS characteristics, upper Hessenberg matrix, Diectionlet transform
PDF Full Text Request
Related items