摘要
隨著科技的進步，網際網路的發展越來越普及，人們對資訊傳輸的質和量有大幅的提升和改善。當數位資訊在網路上傳遞時，如果資訊沒有經過額外的加密進行傳輸，有心人士可以輕易地非法複製檔案藉由網路傳播出去，這對於合法擁有者是一種侵權的行為。因此，科學家提出加入數位浮水印在資訊裡，藉由將數位浮水印藏在多媒體中(像是影片、音訊、或圖片)，對於欲嵌入的資訊達到版權所有保護的效果。 本論文針對不可識別的浮水印嵌入在轉換域上，對於常見的影像攻擊(例如影像切割、影像加入高斯雜訊、或是影像經過濾波器…等)仍然具有強健性。在本論文第四章的部份，著重介紹分數隨機轉換，並且將浮水印嵌入在此轉換域上，並且和其它轉換域(像是離散傅立葉轉換、離散餘弦轉換)做一個客觀的比較。此外，針對目前已存在的方法，提出將浮水印嵌入在振幅的改善方法。
並列摘要
As the technology improves, using the internet becomes more and more popular. We have made a big upgrading and improvement of the information quality and quantity transferred on the internet. When the digital data are communicated on the internet, hackers can easily and illegally copy the data and transfer them through the internet. If the data were not encrypted with extra protection, this will result in a violation of intellectual copyright for legal owners. Therefore, scientists proposed methods of adding digital watermarks to the data. Hiding the digital watermark in the multimedia (like videos, sounds, or pictures) can have a copyright protection. This thesis studies invisible watermark embedding in the transform domain, which is a robust way to embed the watermark in the transform domain for resisting normal image attacks (such as image cropping, Gaussian noise addition, or filtering... etc). In the thesis, we focus on discussions of the discrete fractional random transform (DFRNT) and embedding the watermark in the DFRNT domain, which has better robustness as compare to other transforms (like discrete Fourier transform, or discrete cosine transform). In the Chapter 4 of the thesis, we discuss a lot of experiments with results and also propose a modified method to embed watermark in amplitudes of the transform domain.