Geometric Attacks Resistant Zero-watermarking for Color Images Using Local Multi-channel Fractional-order Exponential Moments and SPM Chaos Mapping

Zero-watermarking is a technique that does not alter the original content for copyright protection and is a popular research area in the domain of digital watermarking. To overcome the limitations of existing zero-watermarking schemes in terms of robustness and security, this paper presents a novel color image multiple zero-watermarking scheme that uses local feature regions (LFRs) and Multichannel fractional-order exponential moments (MFrEMs). The proposed scheme first applies the Scale-invariant feature transform (SIFT) detector to obtain the stable key points of color carrier images and then uses feature scales to build stable LFRs, then MFrEMs of LFRs are computed separately to construct multiple feature sequences using magnitude. SPM chaos mapping is applied to scramble the binary watermarked images to increase the security of the scheme. The feature sequences of each LFR and the encrypted watermarked images are subjected to an exclusive OR (XOR) operation to generate multiple zero-watermarking. Various experiments are performed to assess the proposed multiple zero-watermarking scheme, the results obtained show that the proposed scheme not only has better security but also shows greater robustness against both image processing and desynchronization attacks.