Robust Watermarking Algorithm

With the rapid development and popular application of internet, digitalization has prevailed in every corner of our daily life and brought great convenience for our life, but it’s vulnerable to counterfeit, replication, and infringement, illegal tampering and other attacks. Therefore, digital copyright protection and information security in the network has increasingly appeared to much attention, and become an imminent problem. The development of digital watermarking technology provides an effective way for information hiding、copyright protection and information tamper detection.Digital watermarking is a technique that hides the symbolic information into digital contents. The embedded information constitutes the watermark that is not apparent to the consumer, but can be extracted when required. The watermarking technique has been used in copyright protection of various multimedia contents. Most efforts on watermarking have been concentrated on various media data types such as document, image, audio-. video and so on. With the rapid development,digital watermarking has gotten the great attention and interest of scholars both at home and abroad, became a new research focus in the field of information security.Based on the extensively researching the systems and algorithms for digital watermarking, we propose three novel algorithm. The main contributions are listed as follows:1. To eliminate the limit that the recent zero-watermarking algorithms are sensitive to the variations of illumination, orientation, position and scaling in images, a novel zero-watermarking algorithm is proposed. In the method, images are decomposed into a sequence of binary images using pulse-coupled neural network (M-PCNN), and then the entropies of the decomposed binary images are calculated and regarded as watermark features. A lot of experimental results show that the proposed approach has the Strong robustness to the variations illumination and RST watermarking attacks. It is also shown that the zero-watermark in this method can describe the global characters of the image with low dimensions in comparison with the recent zero-watermarking methods.2. Combining BEMD decomposition and SVD transform, this paper proposes a blind digital watermarking algorithm based on data fusion. This algorithm executes BEMD decomposition on each block of the original image to get IMF6s, and executes SVD transform on each4×4block of the IMF6. The quantization step Q is used to quantify the max singular value in order to embed the watermark in the result of the quantization, multiple copies of digital watermarking are embedded into the carrier image. Extracted copies are fused by image fusion algorithm so as to improve the robustness of the algorithm. Simulation experiments show that the scheme is effective, and achieves higher transparency and strong robustness to common image processing attacks.3. Unlike the existing feature point-based methods which determine and normalize local regions in the host image for embedding and extracting watermark based on the feature points, we developed a new image calibration approach based on the control grid interpolation and polynomial convolution. In order to improve the robustness to the common signal processing methods and the local geometric attack shearing, a novel adaptive wavelet lifting scheme is constructed, and the host image is divided into four non-overlapped and equal-sized blocks, three copies of the digital watermark are embedded into each block using the adaptive wavelet lifting scheme. In the watermark extraction, the image to be detected is calibrated using the image calibration approach based on the feature points extracting from the host image and the detected image by the SIFT,12copies of the digital watermark are extracted from the calibrated image using the wavelet lifting scheme, the watermark is determined by fusing these copies. Experimental results show that the proposed watermarking method yields high robustness to the global geometric attacks and the common signal processing methods. Moreover, it is also robust against the local geometric attacks such as random bend attack and shearing.