| Watermarking techniques consider multimedia data as a communication channel transmitting covert information for a variety of purposes. One widely used technique is the Least Significant Bit (LSB) substitution in the spatial domain, in which an ordinary message, or a watermark, can be imperceptibly inserted in some innocuous looking digital image. A classical LSB scheme makes it computationally simple and easy to hide a large amount of data within a host image. However, such a scheme is fragile in a competitive environment such as the Internet, where adversaries have an incentive to obliterate the embedded watermark. In addition, tradeoffs exist between the capacity, invisibility and robustness of the secret watermark. The possibility of using the intrinsic nature of cover images is explored in this thesis, and a new LSB-based watermarking approach to grayscale images is developed. The proposed method offers better performance in terms of the three attributes through an image-adapting coding. The effectiveness of the method is verified through analysis. The experimental results show that the proposed method improves the robustness of hidden watermarks without compromising their invisibility and capacity. |
