Research On Zero Watermarking Algorithm Against Geometric Attacks

Aiming at the problems of false alarm and high time complexity,this section of paper proposed a fast zero-watermarking algorithm with both the double encryption and strong robustness based on matrix Schur decomposition.Firstly,the algorithm decomposed the original image low-frequency block by discrete wavelet transformation and matrix Schur decomposition to obtain a stable value.Then,it extracted the absolute value of the maximum energy element in diagonal elements of Schur decomposition on the block-triangular matrix.Next,to obtain the characteristic matrix,the average value of the matrix were compared with each element value.In addition,the chaotic map-encrypted feature matrix and the Fibonacci transform-encrypted watermark information were operated to obtain a zero-watermark.Finally,the registration was completed into the certification center(IPR).Experiments show that the similarity between different carriers is less than 0.5,which can effectively solve the high false alarm problem.And compared with the robust zero watermark based on integer wavelet transform,the time required for the algorithm to build zero watermark is increased by 34.48%.Moreover,based on the above research,in order to solve the problems of the zero-watermarking algorithms with weak resistance to geometric attacks,this section of paper proposed a zero-watermarking algorithm against geometric attacks.Firstly,this algorithm used the scale-invariant feature transform to correct the image pixels distortions after rotation correction and determined the near-destructive maximum inscribed square and safe region with approximately pixels lossless in the center.Then,it performed two-level redundant discrete wavelet transformation,and blocked the extracted low-frequency region.Next,it extracted the absolute value of the maximum energy element in the diagonal elements of the Schur decomposition on the blocked triangular matrix to construct a transition matrix.What's more,to obtain a characteristic matrix,it compared the value of each element in the transition matrix with its average value.Finally,using the watermark image and the characteristic matrix constructed a zero watermark.Experimental results show that the robustness against rotation attack is up by an average of 13.14%,compared with the SIFT feature point matching correction algorithm alone.Compared with the GH algorithm and the pseudo-Zernike moment algorithm,the anti-rotation attack robustness is improved by 1.1% and0.93% respectively.Also,it has strong robustness against conventional attacks,scaling attacks,and cyclic shifts and small-range shear attacks.