Illustration Watermarks for Vector Graphics
PG '03 Proceedings of the 11th Pacific Conference on Computer Graphics and Applications
Image annotation watermarking: nested object embedding using hypergraph model
MM&Sec '06 Proceedings of the 8th workshop on Multimedia and security
| Hi-index | 0.00 |
A specific application for image watermarking is constituted by annotation watermarking (sometimes also called caption or illustration watermarking). In this domain, supplementary information is embedded directly in the media, so that additional information is intrinsically linked to media content and does not get separated from the media by non-malicious processing steps such as image cropping or compression. Recently, nested object annotation watermarking (NOAWM) has been introduced as a specialization in annotation watermarking, whereby hierarchical object relations are embedded in photographic images. Earlier work has suggested two methods for achieving this, whereby two main deficits can be identified for those: firstly both methods perform approximation of the user-generated, free-hand, polygonal shape annotations to rectangular annotation areas, which leads to a significant loss in precision of the annotation region after detection. Secondly, both of the previous algorithms do not introduce any security mechanisms, i.e. the object annotations are publicly accessible with no access control mechanism. This paper presents an extension to one of the two previous approaches to suggest a first solution to these shortcomings. For this purpose, a novel method for embedding both the shape of the embedding region and the textual annotations in the phase coefficients of the Discrete-Fourier Transform (DFT) is presented. In order to achieve user dependent access control and improved synchronization for this scheme, the new method involves key-dependency. The key is used to provide an application level security feature and it is not designed yet in the sense of watermark key-space or sub-space security. The suggested methods have been prototypically implemented and first experimental results are presented in comparison to the original method with respect to transparency and robustness. We show that our new scheme may significantly reduce the transparency in terms of Peak-Signal-to-Noise Ratio (PSNR), whereas the overall watermark detection rate decreases, mainly due to local synchronization limitations.