Research Of Text Watermarking Algorithm To Resist Printing And Scanning

With the rapid development of Internet and digital products, awareness of copyright and copyright protection has been growing concern. As an important method of digital product authentication and copyright protection, Digital watermarking technology gets more and more attention and research. Most digital products(documents, books, contracts, etc.) are saved and transferred in text form after printed and as one part of people to communicate and exchange, text document contains rich information. In this paper, anti-print-scan watermarking algorithm is studied in depth.Firstly, this paper introduces the basic principles of digital watermarking system and framework, outlining the basic features and performance assessment criteria of digital watermark technology. Then, according to the characteristics of text images, classify text watermarking technique and discuss the difficulties existing in the process of technological development and explore new ideas in text watermarking algorithm. Finally the quality effect of image by printing and scanning is discussed in detail to provide a theoretical basis for the anti-print scanned text watermarking algorithm research and design.Secondly, text watermarking algorithm based on the pixel flipped is researched. Common binary text image pixel flip models are introduced and an improved method is proposed. On this basis we introduce and realize an anti-print scanned watermarking algorithm based on pixel flipping in line. Compared with the traditional text watermarking algorithm by experiment, this algorithm not only has greatly improvement on the visual effect of the text image after embedding the watermark, but also has bigger promotion in the capacity of watermark.Finally, as the capacity of traditional text watermarking algorithm is small, we research and design text watermarking algorithm based on the pixel flipped in blocks and then put forward an improved method to improve embedding capacity and improve the image visual effect after embedding the watermark information in the premise of ensuring the effect of the watermark extraction. In this paper, we use block coverageilter effective carrier block, then we get the feature matrix of each block in the image and have DCT transform on the feature matrix. The number of pixels of each block which need to be flipped is determined by modifying the DCT high frequency coefficient, then the pixels are flipped and the process of watermark embedding is completed. Experiments show that zhe1 improved algorithm has a larger watermark capacity and can resist printing and scanning.Finally, as the maximum capacity of a single text image is less than the capacity of watermark image, we propose embedded watermark information based on the idea of multiple continuous text images and design text watermark embedding and extraction system. Its feasibility is verified through experiments.