In this paper, we present a new approach for data hiding in digital images by using multiple embedded stages. The new scheme involves feature point detector that orient for a novel Contourlet transform-hiding algorithm. By this way, our model can withstand geometric, desynchronization attacks and other common attacks with a high capacity. | Journal of Science and Technology Volume 48, Issue 4, 2010 pp. 91-98 A GEOMETRICAL ROBUST IMAGE DATA HIDING SCHEME BASED ON CONTOURLET TRANSFORM DUC M. DUONG, DUC A. DUONG ABSTRACT In this paper, we present a new approach for data hiding in digital images by using multiple embedded stages. The new scheme involves feature point detector that orient for a novel Contourlet transform-hiding algorithm. By this way, our model can withstand geometric, desynchronization attacks and other common attacks with a high capacity. The resulting watermarking scheme is suitable for public watermarking applications, where the original image is not available for watermark extraction. Experimental results demonstrate that our proposed technique is robust to a wide range of attacks. Keywords. data hiding, geometrical attacks, watermarking Contourlet transform, 1. INTRODUCTION One of the chief difficulties in watermarking or data hiding scheme is to make the safe level of robustness resistant desynchronization and other geometric attacks. Until now, the effective ways to resist to geometric attacks is still a challenging problem and a one of the research focuses in the field of watermarking. The main issue is the slight geometric manipulation to the watermarked image such as scaling or rotation, which could significantly reduce the possibility of successful watermark retrieval, providing that the watermarking extractor have no knowledge of the distortion parameters. In another word, geometric distortion can easily introduce synchronization errors into the watermark extracting process. In [10], we got significant results in payload, signal processing and some geometric attacks such as cropping and removing. However, our scheme still failed in most of geometric and affine transformations. In this proposal, we use content of the image to achieve synchronization and a Contourlet embedding scheme to embed the message. Firstly, we use Harris detector to obtain feature points with .
