Robust copyright protection scheme for medical images based on dwt, svd, and secret sharing techniques

There are various fine features in medical images, the critical information of which may be distorted when traditional watermarking methods are employed. Even small changes will falsify fine features and therefore influence the final decisions of medical professionals. In this paper, a robust copyright protection scheme that combines discrete wavelet transform (DWT), singular value decomposition (SVD), and secret sharing (SS) techniques is proposed. The proposed scheme employs DWT and SVD to extract important features from medical images. First, a random selecting method is employed to construct a feature array from each feature. With the SS technique, an ownership share can be generated from the feature array, a cover image, and a watermark. The ownership share is a meaningful image that appears just as the cover image. In the identification phase, a master share can be produced from a copyrighted medical image by using DWT and SVD. Next, a decision image is obtained by performing a pixel-by-pixel exclusive-or (XOR) operation between the ownership and master shares. By utilizing an absolute majority of the pixel value for each non-overlapping 2 * 2 block of the decision image to decide a pixel value, an extracted watermark can be created. As a result, the quality of the medical image is unchanged when the image is being protected. Meanwhile, the appearance of the generated ownership share contains meaningful information, thereby facilitating its management. These characteristics allow the proposed scheme to yield performance that is superior to existing schemes. Finally, the experimental results demonstrated that the proposed scheme can resist common image-processing attacks, and that it is reasonably robust as a watermarking scheme. © 2019, National Taiwan University of Science and Technology. All rights reserved.