A Study Of Color Image Watermarking To Resist Cropping Attack

In recent years, digital watermarking technology has been widely used in many fields, such as broadcast monitoring, equipment control, operational tracking, content authentication, identification of the owner, the copy control and ownership verification. In order to embed flag information into the carrier data, the information hiding technology is used in digital watermarking technology. In general, the embedded watermark information for human senses is imperceptible, and only through specific watermark detection or watermark extraction system, we can determine the watermark information exists or not. Though digital watermarking technology has made rapid development, but it also faces many challenges.Cropping attack is a challenge to the digital watermarking technology, which is to cut the carrier with watermarking, it can make the watermark detection algorithm for the carrier failure, because after cropping attack, we can not determine the watermark information from the remaining part, or because the watermark information of the remaining part is so small that the watermark information we extracted from the remaining part is distortion. Therefore, it is an important problem in the research of the digital watermarking technology to resist the cropping attack.The main object of this study is color image in this paper and we propose two digital watermarking algorithms to resist cropping attack. The algorithm is also robust to other attacks, the watermark embedding capacity and invisibility have been improved at the same time. the advantage of color image is the multi-channel, and it is used in the two algorithms, we embed the synchronization information and embed watermarks into different color channels, eliminating mutual interference between the two will be eliminated. The main research work is divided into two parts:1) A digital image watermarking algorithm to resist cropping attack based on Hadamard transform has been proposed. The main idea is to block the image, the same synchronization information is embedded in each block. Firstly we make the image block to 8×8 Hadamard transform, and randomly select Hadamard transform blocks of N, sync information is embedded in the N image blocks; Then the watermark information is obtained by scrambling the watermark image. Finally the watermark is embedded into the intermediate frequency coefficient of Hadamard transform. After cropping, we can find a complete image block by detecting synchronization information, and the complete watermark information can be extracted form the complete image block.2) Based on patchwork algorithm and DCT transform, we proposed a resist cropping attack image watermarking algorithm. Firstly we block the image, in each sub-block, we randomly selected points, in order to embed the synchronization information; Then we use the patchwork method to modify the points selected; and then we embed the watermark information into intermediate frequency of DCT coefficients. Even though there is not a complete image segmentation in the cropping image, the image can be mosaic into a complete image, Then we can extract watermark information from it.The experimental results show that the two algorithms have good robustness and invisibility, and the ability of resisting cropping attack is good, the two algorithms also can resist the JPEG compression, salt and pepper noise, Gaussian white noise.