Adaptive Threshold Based Robust Watermark Detection Method

Abstract

In this paper, we propose a novel blind watermark detector based on channel estimation and adaptive threshold scheme to enhance the robustness of watermarking system. There are channel estimation based blind detectors in the literature. These detectors use reference watermarks to estimate degradations and information watermarks to send the hidden data. As in all estimation problems, however, unreliable estimates may deteriorate detection capability of the system. We employ adaptive threshold scheme to remedy these problems. First, we estimate the raw reliability scores of the watermark channels by using reference watermarks. Then, we transform the raw reliability scores by employing min-max method and logistic function respectively. Hence, we get the channel parameters of the watermark channels. Finally, the information watermark is recovered by using the extracted watermarks and their corresponding channel parameters. The simulations demonstrate that the proposed detector is more robust than existing detectors against various degradations and malicious attacks.