Robust Watermarking using Arnold and Hybrid Transform in Medical Images

摘要

The number of medical digital images continues to grow, thus requiring the protection of patient privacy. It is the reason watermarking on medical images has received special attention. This study proposes a watermarking scheme on the image using the Arnold cat map and hybrid transform (Fast Discrete Curvelet Transforms (FDCuT), Discrete Cosine Transform (DCT), Singular Value Decomposition (SVD)). The Arnold Cat Map technique is applied to pre-processing to increase security and reduce the likelihood of targeted attacks. Experiments were carried out on six modalities: MRI image, x-Ray, CT scan, ultrasound, eye image, and fundus scan. The experimental results show that the watermarked image produces excellent imperceptibility with a peak signal-to-noise ratio (PSNR) above 50 dB, and the structural similarity index (SSIM) is close to 1. The value of bit error rate (BER) 0 and normalized correlation (NC) 1 are obtained in conditions without attack. This technique is resistant to various attacks, namely JPEG compression, adding noise, and filtering. Therefore, the proposed watermarking scheme has good imperceptibility and robustness to be applied in e-health applications.