| The widely using of Internet in current society makes the information being distributed and acquired much more easily. In recent years people pay more attention to the digital works'copyright protection. In the condition that traditional security method can not work well in this field, Digital watermarking is considered as an efficient way to solve this problem. At present, these are a lot of researches in this domain, because of the especial structure of text, the watermarking research on text is much less than the researches on image, audio and video. Furthermore, text watermarking research more focuses on English text. There is few researches on Chinese text.In this thesis, we propose a blind text watermarking algorithm for Chinese texts by utilizing fuzzy clustering theory and Chinese character's properties. The proposed algorithm for text watermarking achieves good balance between the watermark's imperceptibility, capacity of embedding and robustness. On the basis of summarizing the theories which are related to the Chinese character, constructed the Chinese characters'property vector, which is composed of three elements: strokes,frequency and structure types of Chinese characters. Considering the characteristic of Human Visual System, we classify the Chinese characters based on the set of Chinese character's property vector, choose the Chinese characters of which have few strokes and more frequencies, then embedded watermarks through modify its font color attribute value. The improved algorithm can insure the capacity of embedding and imperceptibility. When embedding, a self-correction mechanism is adopted. While extracting watermark, original text is unused. In the process of extracting watermark, it can correct the watermark with proper policy when the integrity of the watermark is corrupted. At last, experiments on this algorithm are implemented and experimental results show that the proposed algorithm has better robustness, transparency and larger capacity. |
PDF Full Text Request