Optimality in Distortion Control in Reversible Watermarking Using Genetic Algorithms

Abstract

This paper proposes a reversible contrast mapping (RCM)-based reversible watermarking (RW) algorithm (RCM-RW) where optimal distortion thresholds adaptive to the image characteristics are used. A generalized form of RCM is developed using a set of transformation functions; each one may be considered as a point on a straight line, called here as operating point. Each operating point offers trade-off benefits on embedding distortion, data hiding capacity and security for the hidden data. The choice of an operating point is governed by adaptive distortion control thresholds, the values depend on the partitioning of the images on smooth, texture and edge regions. However, region-specific optimal distortion threshold is difficult to represent in closed-form expression. Genetic algorithms (GAs), due to their complex searching nature, are used here for calculating the set of distortion thresholds. Simulation results show that high embedding capacity, improved security and imperceptibility of the hidden information can be met simultaneously using a particular combination of such operating points on optimal basis. Extensive simulation results show that the proposed method outperforms the existing RW techniques.