Adapted Quantization Index Modulation for Database Watermarking
Résumé
In this paper, we adapt the robust Quantization Index Modulation (QIM) originally proposed by Chen and Wornell for images to the protection of relational databases. In order to embed a bit of the watermark, we modulate the relative angular position of the circular histogram center of mass of one numerical attribute. We experimentally verify the suitability of our scheme in terms of distortion, robustness and complexity within the framework of one medical database of more than one half million of inpatient hospital stay records. Our solution is very robust against most common database attacks (tuple deletion and insertion and attribute value modications), and the embedded message can be error free retrieved even after a suppression or addition of 80% of the database tuples, making it suitable for database copyright protection, owner identication and traitor tracing.