Research On Digital Image Watermarking And Restoration Algorithm

With the fast development of the information age, multimedia technology has spread into every respect of our daily life. As an indispensable communication tool and information carrier, digital image plays a vital role in our work and life. While during the Internet transmission process, some copyright related digital products are vulnerable to illegal piracy or easily get damaged due to many factors, like hard equipments or human disturbance, these damaged images should be restored or they may affect subsequent operations. Therefore, how to embed the watermark effectively so that copyright can be well protected, and how to recover the damaged images have much research value,On the base of brief introduction about the current digital image watermarking and restoration algorithms, this paper mainly studies robust digital watermarking and color image restoration, the main work is presented as below:1. A digital image watermarking algorithm based on Contourlet transform and singular value decomposition is proposed.Combining the current digital watermarking algorithms, we propose a robust watermarking scheme which combines singular value decomposition,2D-DCT and Contourlet, a more descriptive2D image analysis tool. The original watermark is firstly encrypted with Arnold, and then it is embedded into the cover image twice. The first embedding stage belongs to local embedding, dividing the cover image into blocks and then SVD is applied to each block, the watermark is embedded through modifying the first singular value of each small block. In the second stage, a4-level Contourlet transform is directly applied to the cover image, then the low frequency coefficients are processed by2D-DCT, the watermark is embedded by modifying the DC coefficients. Experiments results show that the proposed watermarking algorithm can achieve good embedding effect and is robust to some regular digital attacks.2. A color image restoration algorithm is presented based on color vector angle.The current restoration algorithms are mainly aimed at gray-scale damaged images, since the gray-scale images use only single color channel to denote, while color images comprise of three color channels, that is R, G and B, if algorithms aimed at gray-scale images are directly applied to color damaged images, then the results won’t be good because of the neglect of color characteristics. Therefore, we proposed a new watermarking method on the basis of Criminisi algorithm, which exploiting the color vector angle to detect and extract the edges so that the restoration sequence can be conducted along the edges direction as much as possible. At the same time, redefining the calculation of priority, and a secondary restoration idea is adopted during the search process of best-matched blocks. SSD principle is used to repair most of the destroyed points in the first step, then a secondary restoration with different windows of variable sizes or directions is then employed to those who do not meet the defined standards in the first stage;. After parameters discussion and a number of experiments, the proposed algorithm is proved to be capable of restoring color damaged areas and can achieve good repairing results.