Wave-atoms-based multipurpose scheme via perceptual image hashing and watermarking

This paper presents a novel multipurpose scheme for content-based image authentication and copyright protection using a perceptual image hashing and watermarking strategy based on a wave atom transform. The wave atom transform is expected to outperform other transforms because it gains sparser expansion and better representation for texture than other traditional transforms, such as wavelet and curvelet transforms. Images are decomposed into multiscale bands with a number of tilings using the wave atom transform. Perceptual hashes are then extracted from the features of tiling in the third scale band for the purpose of content-based authentication; simultaneously, part of the selected hashes are designed as watermarks, which are embedded into the original images for the purpose of copyright protection. The experimental results demonstrate that the proposed scheme shows great performance in content-based authentication by distinguishing the maliciously attacked images from the nonmaliciously attacked images. Moreover, watermarks extracted from the proposed scheme also achieve high robustness against common malicious and nonmalicious image-processing attacks, which provides excellent copyright protection for images.