Optimal capacity threshold for reversible watermarking using score function

Abstract

Histogram shifting is an important technique of reversible watermarking, which can embed large payloads into digital images with low distortion. The technique must determine two threshold values to achieve the lowest possible distortion. Appropriate threshold values might be found by iterative methods, but it is computationally inefficient when the payloads are high and varied. In this paper, we show that the optimal threshold values lie on a straight line and occur at the boundary of the payload-satisfying region. Moreover, we propose a high performance algorithm to approximate the optimal threshold values. Under the same image quality, experimental results indicate that the proposed scheme could get closer threshold values to the optimal threshold values, compared to previous work. Therefore, it requires a smaller number of iterations to obtain the desirable threshold values.