A new approach for relational database watermarking using image

Proving ownership rights on outsourced databases is a vital issue. This paper proposed an improved algorithm for database watermarking. In the experiment, we embed an image watermark into numerical attributes in a relational database. It refers to image elements as 2-dimention matrix elements and tries to embed medium size image in small scale relation. By applying this method we can obtain upper Correction Factor for image. Furthermore at the phase of embedding watermark, this algorithm is capable of minimizing data variation method which assures the minimum modification to original database without decreasing imperceptibility. In extracting process majority voting method is employed to get retrieve the correct watermark. To improve this process, recovering step is added to the extracting process. This step tries to recover extracted image and guess missing elements value, so the watermark can be detected even in a small subset of a watermarked database. Detecting the watermark neither requires the original database nor the watermark.