The Research On The Digital Image Watermarking Algorithm To Resist Geometric Distortion

With the popularity of digital information and multimedia networks, digital watermarking technology has played an important role in protecting the copyright of digital products. At present, the most of watermarking algorithms increase embedded strength of the watermarking by using the spread spectrum, the channel coding and the performance of visual system. These algorithms are robust to the commonly signal processing such as compression and noise, but they are difficult to resist the unsynchronized attacks such as rotation, scale and translation. Therefore, how to solve problem of the synchronization in embedding and detecting is a still popular research.In this paper, the research of the digital image watermarking is based on the wavelet domain. Before the watermark detection, the stable SIFT(scale invariant feature transform)feature points are used to estimate distorted parameters of the watermarked image possibly distortedly, then watermarked image is corrected and restored. Therefore, resisting geometric distortion is solved in the watermarked image. Two different algorithms are designed in grey image and color image based on the feature points corrected.(1) In the grey image, the binary image watermarking is embedded into the second level of the low wavelet frequency in the grey image; Before watermark detection, the stable SIFT feature points of original image and watermarked image are extracted, then these points are used to estimate geometric distortion parameters and watermarked image is corrected and restored. So the algorithm ensures the accuracy of the extracted information, and anti-geometric distortion is achieved in the grey image.(2)In the color image, compared the relations of the wavelet coefficients between the blue and green components of the image, the binary image watermarking is embedded into the second level of the low wavelet frequency in the blue component. The SIFT feature points of red component are used to estimate geometric parameters and distortedly watermarked image is corrected, then non-distorted watermark is detected. Thus anti-geometric distortion is achieved in the color image.Watermarking synchronization is very important to robustness. The SIFT feature points are used as a template to correct geometric distortion based on the scale invariance. So watermarking detection is robust to rotation, scale, translation and brightness change, and the watermarking algorithm guarantees the anti-geometr- ic distortion.The paper introduces the domestic and foreign development of the digital watermarking, and presents designing detail of the algorithms, then shows the algorithms are effective through a series of digital watermarking experiments. A new anti-geometric way of image watermarking technology is presented.