LPM-based RST invariant digital image watermarking

Abstract

Based on log-polar mapping (LPM), this paper presents a novel digital image watermarking scheme that is invariant to rotation, scaling, and translation (RST). We embed watermark in the log-polar mapping (LPM) of the Fourier magnitude spectrum of original image to simplify the effects of RST transformations into simple shifts. For watermark embedding, the approximate ILPM is employed to replace ILPM, in order to eliminate the imprecision caused by ILPM. Therefore actually watermarks are embedded in the Fourier magnitude spectrum of the original image, to achieve the effect of being embedded in LPM domain. Watermark extraction does not need original image, we use exhaustive search in the embedding area to handle the shift of watermark positions in LPM domain caused by rotation and scaling in spatial domain. The scheme avoids computing inverse log-polar mapping (ILPM) to preserve the image quality. Evaluations demonstrate that the scheme is invariant to rotation and translation, sufficiently invariant to scaling transformation, and very robust to JPEG compression and other attacks.