The Research Of The Document Tracking System Based On Watermarking Technology

With the development of The leakage of confidential documents by means of printing and copying is an emerging problem of intranet security,however,the existing document security solutions are mostly based on the control of access and use permissions for digital form documents.Therefore,it is impossible to track or trace printed documents in analog simulation form.In order to solve this problem,this paper researched and implemented a document tracking system based on digital watermarking technology.Considering the good transparency,strong robustness and high watermark capacity of digital watermarking technology,this paper has chose the digital watermark technology to trace the source of scanned and printed documents.The key point is that based on the word spacing coding algorithm,the paper has presented a dynamic group spacing text watermarking algorithm.Primarily,the algorithm dynamically groups the text lines and pairs them together according to the location of the punctuation.Then the algorithm adjusts the size relationship of character spacing within the paired packets,to make it consistent with the watermark information,and finally realize the watermark embedding.This algorithm uses group spacing instead of word spacing to encode,reducing the amount of adjustment eventually assigned to each word spacing and improving the visual effect of the document images containing watermarks.Meanwhile,the algorithm takes punctuations as the boundary,naturally and dynamically segmenting the entire text line,which has avoided the problem of peripheral spacing abnormity of punctuations which is easily caused by the direct adjustment of word spacing.Simulation results have showed that the dynamic group spacing text watermarking algorithm had good transparency and could resist multiple attacks such as printing or scanning attacks.The projection segmentation method can not correctly segment the Chinese characters whose components are not connected,such as these with left-middle-right structure.Hence aiming at solving this problem,a Chinese character segmentation algorithm based on machine learning is proposed in this paper.Using SVM to learn multiple features such as character connectivity,character template similarity coefficient,etc,this study can realize accurate recognition of Chinese characters.In the aspect of paper design,a printing tracking system has been implemented.Deployed on a controlled terminal,the tracking system client is responsible for automatically and forcibly embedding end-user identity into its printed document.Among them,a customized print processor has been used to redirect the printing task of the controlled terminal,so as to achieve the printing interception.Using virtual printing technology to realize the conversion of documents to images,and applying OpenCV to actualize the watermark embedding and the extraction algorithm,are two other key technical points in this paper.Otherwise,the management end has been designed and implemented by the use of the MFC development framework of Windows platform,achieving the management of department information,user identity information and watermark,as well as the watermark extraction of problem document.Finally,the test of the tracking system’s functionality and performance has proved that the system could effectively locate the person responsible for printing leakage.