Research On Text Watermarking Algorithm Based On The Fuzzy Clustering Theory

With the rapid development of computer and network, human walks into theinformation age. Human can receive various media forms of information such as video,images, sounds, and text every moment. in the exact expression, contains large amountof information and flexible reading, the characteristics of the rapid retrieval, Textbecome the most widely used media forms of information because it can exactexpression, contains large amount of information, flexible reading and rapid retrieval.For example, legal documents, the government official documents, electronic document,network news, BBS and blog all are text. Text has become the new important carrier ofhuman communication. With the wide use of text, its copyright protection is becomingmore and more important.Text digital watermarking technology is one of the ways to solve this problem. Thetechnology that embedded copyright information in the text file to confirm theownership of the copyright evolved. Text watermark results relatively small becausetext file has the characteristics of single structure and special processing method.Watermark can be embedded in text files, so secret information can be embedded in textfiles too. This involves more important problem than copyright confirm--nationalsecurity. First, we can use the text digital watermarking technology to review suspicioustext files, intercepted and crack secret information come from outside timely, stop theinformation handled through this channel from home to the relay. Second, we can usethis technology to transmit confidential information. Therefore, the study of text digitalwatermark is particularly urgent and important.This paper establishes a clustering model using Chinese characteristics in textdocument as its variable, based on fuzzy clustering theory. The algorithm chooses theChinese characters of less stroke number, higher frequency and higher distributionuniformity compose feature vector as the original data of fuzzy clustering analysis. In atext, we select higher frequency Chinese characters in order to improve the capacity ofthe watermark, select higher distribution uniformity Chinese characters in order toreduce watermark’s attack damage, and select less stroke number Chinese characters, itis relatively thinner and has good visual not visible when fine-tuning its character width.Before watermarking embed, the original image coding use m series pseudo randomsort to scrambling. Then, we embedded watermark coding by fine-tuning Chinese character width and every16coding characters Tags for a group by fine-tuning theChinese characters RGB value. The coding has certain autonomous error correctionfunction. When the carrier Chinese characters lost part the watermark can be fullyrestore. We respectively set the judgment of the watermark extraction by thewatermark’s various attack ways, it make the watermark can fully recover in somedamage condition and partly recover in other damage condition. Experimental resultsdemonstrate the effectiveness of the algorithm against partial deleting, adding, andformat adjusting of text. The algorithm has good concealment, well no visibility, andbigger capacity.According to human visual system, it classifies the Chinese characters in textdocuments by their characters of stroke number, frequency and distribution uniformity.Then it chooses the Chinese characters of less stroke number, more frequency andhigher distribution uniformity to be embedded. Experimental results demonstrate theeffectiveness of the algorithm against partial deleting, adding, and format adjusting oftext. The algorithm has good concealment.