An SVM-based robust digital image watermarking against desynchronization attacks

In image watermarking area, the robustness against desynchronization attacks, such as rotation, translation, scaling, row or column removal, cropping, and local random bend, is still one of the most challenging issues. This paper presents a support vector machine (SVM)-based digital image-watermarking scheme, which is robust against a variety of common image-processing attacks and desynchronization attacks. To protect the copyright of a digital image, a signature (a watermark), which is represented by a binary image, is embedded in the digital image. The watermark embedding and watermark extraction issues can be treated as a classification problem involving binary classes. Firstly, a set of training patterns is constructed by employing two image features, which are the sum and variance of some adjacent pixels. This set of training patterns is gathered from a pair of images, an original image and its corresponding watermarked image in the spatial domain. Secondly, a quasi-optimal hyperplane (a binary classifier) can be realized by an SVM, and the SVM can be trained by utilizing the set of training patterns. Finally, the trained SVM is applied to classify a set of testing patterns. Following the results produced by the classifier (the trained SVM), the digital watermark can be embedded and retrieved. Experimental results show that the proposed scheme is invisible and robust against common signals processing such as median filtering, sharpening, noise adding, and JPEG compression, etc., and robust against desynchronization attacks such as rotation, translation, scaling, row or column removal, cropping, and local random bend, etc.