Research And Implementation Of Digital Watermarking For OOXML Format Documents

With the rapid development of Internet technology and big data technology,a large number of image,video,audio and text and other types of data are widely spread on the Internet.The ownership protection of these data has always been a research hotspot.In recent years,digital watermarking technology has been applied,and its application in image,video,audio and other fields has made great progress.However,due to the special properties of text data,it is always a very difficult problem to embed watermark in text while keeping the invisibility,high anti attack and robustness of watermark.Aiming at the widely used OOXML document,two methods are proposed in this paper.Firstly,a robust method of embedding and extracting text watermarks proposed in this paper,which the watermarks are embedded into the text by transforming the RGB color attribute values of <w:r> tags in OOXML document content.The method can get high watermark embedding capacity,and can embed two bits of watermark information between each two <w:r> tags at most.Based on the good characteristics of OOXML format documents,and only the color of the characters is adjusted slightly,the content of the text will not be changed after the watermark embedded,so that the subtle color changes cannot be recognized by the naked eye,which ensures the good invisibility of the watermarking approach.The method can resist "copy and paste in original format","save as",and a small number of "insert" and "delete" operations;in addition,it can detect the location of a small number of "insert" and "delete" operations.This method can be applied to OOXML documents,such as word documents with the suffix of "docx".In view of the fact that the watermarking method based on color attribute value transformation can not deal with the attack of "format clearing",another text watermarking method which combined with natural language proposed.The method uses the "shape" in OOXML format document to be as a separator,in which its attribute values are set-up to make it invisible.It is a method of embedding and extracting watermarks by judging the relationship between adjacent Chinese characters by the word segmentation tools in natural language processing.The proposed method has good invisibility,large watermark embedding capacity and strong robustness,which can resist most of the format attacks,and it can also deal with various attacks on text content.We design and implement the watermark system of the above two methods,and analyze its performance by some experiments to prove the feasibility in practice.Finally,the advantages and disadvantages of the proposed methods are summarized,and the future research contents are prospected.