Robust Wavelet-Based Facial Image Watermarking Against Geometric Attacks Using Coordinate System Recovery

A coordinate system of the original image is established using a facial feature point localization technique. After the original image transformed into a new image with the standard coordinate system, a redundant watermark is adaptively embedded in the discrete wavelet transform(DWT) domain based on the statistical characteristics of the wavelet coefficient block. The coordinate system of watermarked image is reestablished as a calibration system. Regardless of the host image rotated, scaled, or translated(RST), all the geometric attacks are eliminated while the watermarked image is transformed into the standard coordinate system. The proposed watermark detection is a blind detection. Experimental results demonstrate the proposed scheme is robust against common and geometric image processing attacks, particularly its robustness against joint geometric attacks.