In this report, we summarized our investigation of the watermarking techniques based on DCT and DWT as well as the data hiding technique. A set of random numbers can be embedded into the original images as the digital watermark and successfully detected. The major advantage of these watermarking techniques is that the original image is not needed in the detection process. With this property, a single set of watermark can be applied to a huge image database. When these images are widely distributed, the ownership can be easily verified as long as the watermark is available, which contains a very small amount of data compared with the large image database. The performance of the watermarking technology is measured by the invisibility and the robustness. The results indicate these two algorithm are relatively well. The image quality degradation is almost imperceptible to human eyes and the watermark can be detected after some common image processings.

         The data hiding technique is a more natural choice for humans because it can be seen by eyes. In comparison, the digital watermark only needs detection to show its existence, but the data hiding needs to extract the exact hidden information. Such a requirement definitely reduces the flexibility in implementation; hence lower robustness can be expected. The major advantage of this method is that a logo image contains more information than the random numbers set. A partially damaged embedded image may still be sufficient enough to determine the ownership under human's judgement. The results showed that the embedded image remains readable under some common image processings.

         Although these methods have reasonable performance, the robustness is not proved by rigorous mathematical reasoning. There should exist methods that might easily remove or impair the watermark and the embedded image. Especially the data hiding technique we proposed is not secure enough. Malicious attacker can surely remove the mark. However it is hard to achieve robustness and security at the same time. Digital watermarking and data hiding are relatively new research field so there is much can be done. This report is just a start.