| Summary: | 碩士 === 靜宜大學 === 資訊管理學系研究所 === 92 === The rapid technological advances of computers and the Internet have been directly influencing people’s lives. The endless possibilities of the Internet can be attained for a much lower price than the past. This causes issues of ownership as digital information can be easily duplicated without degradation. Digital watermark, which is to hide the owner’ s secret information or the logo, were thus developed to provide a means to a proof of ownership.
Most watermark researches have focused on multi-media files taking advantage of their coding can withstand slight modification without being audio-visually noticed. There has been little research on watermark as applied to document files because one runs the risk of damaging the document’s structure when applying multi-media based watermarking techniques to a document file.
In this thesis, we devise a secret-sharing based watermark model for document files and implement a system for RTF files. We randomly select n letter-segments and convert these segments into small 2-color image files for embedding the n unique “shadows” of the main secret, which are created from the (n, t) secret sharing scheme. In this way, the system preserves the document style while taking advantage of applying data hiding technique of 2-color images. Meanwhile, spreading the secret shadows into the entire document provides the system with the capability of proving the ownership for partial copying.
Our system is secure as the following two mechanisms complicate the analysis: (a), It embeds n unique shadows instead of embedding the same secret n times. (b), Each shadow information is embedded into different non-uniform sections determined by a sequence-key only known to the owner or the authorized user.
|
|---|
