Region-based reversible medical image watermarking algorithm for privacy protection and integrity authentication

Abstract

Medical images are widely used in telemedicine, sharing and electronic transmission between hospitals. While enjoying convenience, medical images also face privacy disclosure, illegal copy and malicious tamper, etc. It is highly important to ensure privacy and integrity of medical images. Cogitating the above needs, a reversible medical image watermarking algorithm for privacy protection and integrity authentication is proposed. Firstly, the medical image is divided into the ROI (Region of Interest) and RONI (Region of Non-Interest) based on active contour model. Then, the proposed “Three-Dimensional Watermarks” are generated, including authentication watermark, restoration watermark and privacy watermark, which are produced by the novel Parallel Lattice Hash Function, the proposed Neighborhood Difference Method, and the proposed encryption algorithm, respectively. Moreover, “Double-Layer Reversible Embedding Strategy Based on Difference Expansion” is modified in ROI to improve embedding capacity, and “Histogram Modification Reversible Embedding Strategy of Difference Image” is modified to adaptively acquire four or more peak points, which is more flexible than common algorithms. Experimental results confirm the efficient of the proposed scheme, and demonstrate it not only realizes privacy protection, integrity authentication, reversibility, but also holds the characteristics of higher security, larger capacity and better restoration quality.