A Novel Multi-logo Image Watermarking Scheme Resisting Geometrical Attacks

In watermarking applications, the robustness of the watermark to geometric manipulations is a critical issue. This is due to the fact that changing the image size or its orientation could make the receiver lost synchronization with original watermarked image. In this paper, a novel image watermarking scheme resisting geometrical attacks is proposed. We don´t need to normalize the original image. The improved normalization procedure, which we deploy in this paper, geometric moments only are used when detecting the watermarks to recover the attacked image, original image being not needed. Therefore, it can save a lot of computation. For watermark security, logo watermarks are scrambled before embedded. Then the original image is decomposed by discrete wavelet transform, and watermarks are embedded into low frequency sub-band. The experimental results show that this scheme can resist many geometrical attacks including cropping and affine transform attacks and has the advantages of good robustness, precise correction, and efficient computation.