Improvement on Information Capacity of Watermarked Images by Using Multi-Valued Area of Embedded Signals

Digital watermarking technology is very useful as a service for embedding ID data to analog content such as an image on printed material. Recently, many algorithms have been proposed for detecting watermarks from a captured image by a camera-equipped cellular phone. This paper focuses on an approach to embedding the ID data in a frequency domain. Although the approach has an advantage of robustness against a partial distortion of the captured image, the previous method along the approach has the problem of low limitation of embedded information capacity. This paper proposes a watermarking method for improving embedded information capacity by using multi-valued area of signals instead of binary signals. The feasibility of the proposed algorithm is confirmed in experiments using actual printed images.