A zero-bit Fourier image watermarking for print-cam process

Smartphone watermarking has many potential applications but also many challenging issues such as being able to withstand print-cam attacks. These include perspective deformations that can strongly deform the freehandedly digitized image. In this paper, we present the design of an image watermarking for print-cam process in the context of an industrial security application for identity images. The method uses watermarking based on the Fourier transform and a specific correction pre-process. These corrections combine frame-based perspective rectification of the freehandedly captured images, a Wiener filter to decrease image blurring and adjustments to reduce color degradations. Results show that the Fourier watermarking method gives better results than state of the art of existing watermarking methods often chosen to cope with print-cam attacks. Fourier watermarking provides a total error rate of 1%, which is compatible with the targeted industrial application. The error of the other methods is at best 25%. Finally, no noticeable difference was found between the two smartphones tested (iPhone 6 and Samsung S5), in terms of error rate, for the proposed method.