Research Of Image Watermarking Algorithm For Resisting Geometric Attacks

Digital watermarking technique is an effective method for digital content protection such as copyright protection and becomes an important research area in the field of information hiding. The digital production embedded watermark is apt to suffer attack; one of the key problems of watermarking is the robustness to geometric distortion, which can disable watermark by destroying the synchronization between watermark and image. So the research of image watermarking resist to geometric attacks is signification.Many digital watermarks now available for images are sensitive to signal processing and geometric distortions, a robust digital image watermarking algorithm is proposed. Realizing the geometric invariant of watermarking by the geometric invariant of singular value application; after decomposing the image with discrete wavelet transform, then applying the singular value decomposition to the low frequency domain, and embedding the watermark image with chaotic scrambling by modifying the singular values, and embedding the watermark data to the middle frequency coefficients of wavelet domain; We extract watermarks in the intermediate frequency and low frequency region and compare them, using the extracted watermark which is better. The examination algorithm of watermark uses the double watermark comparison examination.Based on the point of geometric correction, an image watermarking algorithm to geometrical attacks is proposed. Based on the theories of invariant centroid, and extracting two invariant centroids as factors to correcting geometric transformation, the watermarked image could be corrected; After decomposing the whole image with 3 level of discrete wavelet transform and transforming the watermark image by Arnold shuffling, then embedding the watermark data to the middle frequency coefficients of wavelet domain according to the conceal quality of Human Visual System.The experimental results demonstrate that the two algorithms are robust to general image processing and geometric attack such as rotation, scaling, translation.