Reversible Watermarking Based on Statistical Quantity Histogram

Abstract

The histogram shifting based reversible watermarking techniques have attracted increasing interests due to their low computational complexity, high visual quality and considerable capacity. However, those methods suffer from unstable performance because they fail to consider the diversity of grayscale histograms for various images. For this purpose, we develop a novel histogram shifting based method by introducing a block statistical quantity (BSQ). The similarity of BSQ distributions for different images reduces the diversity of grayscale histograms and guarantees the stable performance of the proposed method. We also adopt different embedding schemes to prevent the issues of overflow and underflow. Moreover, by selecting the block size, the capacity of the proposed watermarking scheme becomes adjustable. The experimental results of performance comparisons with other existing methods are provided to demonstrate the superiority of the proposed method.