Adaptive multiple zero-watermarking based on local fast quaternion radial harmonic Fourier moments for color medical images

Abstract

The copyright protection of medical images is receiving increasing attention. Zero-watermarking schemes are a critical technology for copyright protection of medical images and show better imperceptibility performance than embedded watermarking schemes. However, existing zero-watermarking schemes still have some limitations that prevent them from fully exploiting their potential, including failure against complex desynchronization attacks and high computational complexity. To this end, we propose a novel zero-watermarking scheme based on fast quaternion radial harmonic Fourier moments (FQRHFMs) that is resistant to desynchronization attacks. Specifically, this article formulates multiple local feature regions (LFRs) based on a speeded-up robust feature (SURF) operator and provides a fast computation method for QRHFMs by using fast Fourier transform (FFT). The magnitudes of FQRHFMs are computed in each incomplete overlapping LFR for generating multiple zero-watermarks to resist desynchronization attacks. Experimental results show that the proposed scheme achieves state-of-the-art robustness against various desynchronization attacks (e.g., translation, cropping, and length-width ratio changing), and outperforms existing local zero-watermarking frameworks with regard to accuracy and efficiency.