CRT-Based Robust Data Hiding Method by Extracting Features in DCT Domain

Abstract

In this paper a robust data hiding method is presented by applying Discrete Cosine Transform (DCT) to the host signal and performing feature extraction, based on Chinese Remainder Theorem (CRT). Improving the robustness reduces the imperceptibility in watermarking systems due to the tradeoff between imperceptibility, robustness and capacity in data hiding schemes. The proposed method offers high imperceptibility alongside robustness. To embed the hidden information CRT based feature extraction process is carried out in DCT domain. Extracted features are distances of CRT remainders. These distances are not varied much in noisy situations thus it is possible to extract the hidden bits with high accuracy. To analyze the robustness of the proposed method common signal processing manipulations are applied to the watermarked images. Bit Correct Rate (BCR) values are calculated for extracted watermarks to evaluate the robustness. Results are compared to other state of the art methods and confirm the superiority of the proposed method.