標題: 以Microsoft Office文件作資訊隱藏之新研究
A Study on New Techniques for Data Hiding via Microsoft Office Documents
作者: 劉宗原
Liu, Tsung-Yuan
蔡文祥
Tsai, Wen-Hsiang
資訊科學與工程研究所
關鍵字: 資訊隱藏;Office文件;追蹤修訂資訊;可逆式可視浮水印;Data hiding;Microsoft Office document;change-tracking information;Lossless visible watermarking
公開日期: 2009
摘要: 數位資訊處理與網際網路技術的快速發展,使資訊隱藏技術的發展愈為重要,其應用也更多元化。目前之研究偏重在影像、聲音、影片等檔案中藏入資訊,但在產官學界經常產生、使用、互通之Microsoft Office文件卻少有人研究探討。該類檔案之格式及特性迥異於影像、聲音、影片等檔案,需要嶄新之方法以達到版權保護、資料驗證及秘密傳輸等目的,極具研究價值。本論文針對Microsoft Office文件探討其特性並提出了六個在Office文件隱藏資訊之研究範圍,包括於Microsoft Office文件之文字中隱藏資訊、於文字編排中隱藏資訊、於嵌入之多媒體物件中隱藏資訊、於嵌入物件編排方式中隱藏資訊、於Microsoft Office文件輔助數據資料中隱藏資訊以及於實體檔案格式中隱藏資訊。本論文亦提出了六種具體的新的資訊隱藏方法及應用,可適用於常見之Microsoft Word、Microsoft Excel、Microsoft PowerPoint以及Microsoft Visio等檔案類型。 首先,本論文針對Microsoft Office 文件可多人編輯之特性提出在Microsoft Word文件中利用追蹤修訂資訊以及賀夫曼編碼(Huffman Coding)技術隱藏秘密之新方法。針對文件內容常被轉載之應用,我們提出多重適用性簽章方法(MUST)以及單樹根簽章方法(TRUST)兩種雜湊值及簽章處理方法並結合資訊隱藏技術以在Word文件中有效的達到轉貼資訊之來源驗證之目的,並提出二維多重適用性簽章方法(2D-MUST)及二維單樹根簽章方法(2D-TRUST)兩種二維雜湊值及簽章處理方法以在Microsoft Excel二維試算表文件中做轉貼表格之來源驗證,以及利用二維多重適用性簽章方法以偵測二維試算表文件內容可能遭竄改之應用。而針對文件內容常被剪貼、複製、收集之特性,本論文提出利用透明字元顏色及依權重統計偽隨機資訊隱藏順序之技術以在PowerPoint等文件中藏入隱密浮水印之新方法以達到來源追蹤等目的。另外,Microsoft Office 文件中常包含各式影像、繪圖等,本論文提出了利用物件群組套疊關係以隱藏資訊以及利用創新的複合式一對一映射理論在影像中嵌入可逆式可視浮水印之新方法,而其中提出的可逆式可視浮水印方法可用於嵌入多種浮水印如單色不透明浮水印以及半透明全彩浮水印等。以上六種方法,皆為創新之作,實驗結果顯示論文提出的方法皆具有可行性及實用性。
With the advancement of digital information processing and Internet technologies, the field of data hiding has become more and more important, and their applications have become more and more diversified. Many techniques have been proposed for hiding data in images, videos, and audios, but there are relatively few researches devoted to data hiding in the popular Microsoft Office documents. Microsoft Office documents are in very different formats and have unique characteristics compared to images, videos, and audios, and so new techniques are needed for embedding data in such media for the purpose of copyright protection, covert communication, authentication, and so on. In this study, we investigate the characteristics of Microsoft Office documents pertaining to data hiding applications and identify six areas for researches of data hiding via such documents: data hiding via texts; data hiding via text formatting and layout; data hiding via multimedia contents; data hiding via multimedia formatting and layout; data hiding via auxiliary data; and data hiding via physical file formats. We also propose six specific new methods and applications for hiding data in Microsoft documents of Word, Excel, PowerPoint, and Visio. First, exploiting the characteristic that documents can be written by multiple authors, a new method is proposed for embedding data in Microsoft Word documents for the purpose of covert communication by using change-tracking information and the Huffman coding technique. Then, to tackle the problem that contents in a document are often cited and included in another document and that there is a need to authenticate the fidelity and source of the cited content, a method is proposed in this study which combines data hiding techniques with two different hash value processing techniques – MUST and TRUST – that can efficiently verify the fidelity of cited contents in a Word document. Furthermore, two two-dimensional hash value processing techniques 2D-MUST and 2D-TRUST are proposed that allow quotations of the form of a two-dimensional table from a Microsoft Excel spreadsheet to be authenticated. Also, the 2D-MUST is demonstrated to allow effective fidelity authentication and modification detection of spreadsheet contents. To address the characteristic that contents within Microsoft Office documents are often moved, copied, and collected together, a new method is proposed for embedding invisible watermarks into slide presentations for the purpose of source tracking by using blank space coloring and weighted voting techniques. Finally, via rich media such as drawings and images contained in Microsoft Office documents, two data hiding methods are proposed, with the first using the different nested grouping relationships of objects to embed information in Microsoft Visio drawings, and the second method using a new generic approach of compound one-to-one mappings to embed completely-removable visible watermarks into images. The latter method was shown to be able to embed opaque monochrome watermarks as well as translucent full color visible watermarks, which is the first in publications to the best of the author’s knowledge. Experimental results are included to demonstrate the feasibility of all the proposed methods.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079123811
http://hdl.handle.net/11536/40318
Appears in Collections:Thesis


Files in This Item:

  1. 381101.pdf