| Summary: | 碩士 === 國立臺灣大學 === 資訊工程學研究所 === 89 === Abstract
Recently, with the development of network technique, Internet has changed human living style. People are used to using computers on Internet for credit card paying, private correspondence, and data communication. Hacker could steal much unprotected information from Internet; therefore, how to protect intellectual property right (IPR) and secret data communication are important. IPR protection, however, needs the coordination of the society, government legislation and computer technique support. We would like to develop computer technique to operate in coordination.
Digital image is one of important data in Internet communication, users often download private images without copyright. “Data hiding” or “steganography” is the digital image protection mechanism developed those years. Data hiding is the art of hiding a message signal in a host signal, host signal is the digital image in our research; people could hide information into digital image for intellectual property protection or secret communication. Different from the cryptography which just hides the contents of information, data hiding concerns the tight combination between watermark and host signal. Hence, even the pirate cracks the watermark system and gets the hidden data, there is nothing to do with owner declaring the ownership of the host signal.
There are three mechanical classifications of data hiding: robust watermark, fragile watermark, and caption-based watermark. We focus on caption-based watermark. Caption-based watermark is the mechanism of hiding data for secret communication; it delivers information behind human visual perception, therefore watermark undetectablility is necessary; besides, caption-based watermark should satisfy high capacity data communication, it is our research topic.
Caption-based watermark mechanism hides data in different host signals. Messages could be hidden in video sequence or image without audience perception. Those messages could be the duration of video, video content, or image ownership information. Capacity of watermark, however, is the research analyzing the characters of image to maximize the embedding watermark length without human perception. In the beginning, there is only single one watermark in image, people judge the ownership by detecting the watermark existence or not. With the development of watermark technique, one bit is not enough because people expect to hide more information like owner name and user copyright, and so on. Robust watermark carries at least sixty-four bits information recently. Meanwhile, secret communication should take more information in order to deliver data, so capacity of data hiding should be large than sixty-four bits. Therefore, in the following work, we would satisfy the high capacity requirement of watermark in frequency domain for secret communication.
|
|---|
