標題: 在文件影像上做資訊影藏之研究
Information Hiding in Document Images
作者: 翁連奕
Lien-Yi Weng
蔡文祥
Wen-Hsiang Tsai
資訊科學與工程研究所
關鍵字: 文件影像;浮水印;版權保護;掃描與列印;影像驗證;資訊影藏;document image;watermarking;copyright protection;print-and-scan;image authentication;information hiding
公開日期: 2004
摘要: 由於數位影像技術的快速發展,數位影像很可能被非法竄改、編輯與複製。因此有必要去發展能保護數位影像的方法。在本論文中,我們針對公文影像提出了數種數位浮水印的技術以達到能夠保護版權並指出影像被竄改之處的目的。針對灰階公文影像,我們利用數位浮水印提出了一種影像驗證的方法,且我們希望即使此影像被列印與掃描後,其驗證資訊還能被準確的抽取出來,同時若影像被破壞後,我們也能指出被破壞的區域。另外我們也提出了一種在公文影像上加入強軔性浮水印的方法,以達到保護版權的目的。針對黑白公文影像,我們提出了一個資訊隱藏的方法,因此可藉由嵌入在影像內的浮水印來保護其版權。最後,我們提出了一套整合性的資訊隱藏的方法來同時藏入浮水印與驗證訊號,其特色為不但可以從影像中抽取出隱藏的資訊,同時若影像遭受到竄改時,我們也可指出影像被竄改之處。相關的實驗結果證明其所提的方法是可行的。
Because of the rapid development of image processing, it often occurs that digital images are duplicated or edited without authorization. As a result, it is desired to develop a scheme to protect the copyright and to verify the integrity of digital images. In this study, we propose several watermarking techniques for digital document images for copyright protection and image authentication. For grayscale document images, a method based on semi-fragile watermarking against print-and-scan operations for image authentication is proposed, which is useful for verifying the fidelity and integrity of document images. Then a method for copyright protection of document images is proposed, which embeds a robust watermark into a document image, based on the use of edge direction histograms with circular interpretation. For color images, a method for image authentication is also proposed. Generation of authentication signals and the positions for embedding them are controlled by two keys. A suspicious image can be verified for tampering proof by comparing the difference between the embedded authentication signals and those generated by the two keys. For binary images, a data hiding method for copyright protection is proposed, which embeds up to three bits in a 4×4 image block by rearranging the black pixels in the pre-selected 2×2 image block in the 4×4 block. A method for integration of watermark and authentication signals is finally proposed. Good experimental results prove the feasibility of the proposed methods.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT009223534
http://hdl.handle.net/11536/76584
Appears in Collections:Thesis


Files in This Item:

  1. 353401.pdf