Digital Watermarking Algorithm Based On The Image Feature Of Document Image

With the rapid development of Internet technology, an increasing number of digital products spread through the network. This is a great convenience to users, but also brought the information hidden. How to take appropriate measures to protect the author's copyright, how to ensure that users receive product information correctness and completeness is an issue. In this context, emerged the digital watermark technology.Among the many types of multimedia data, it is easier for the document images to be copied and disseminated, and in many cases, document images have a higher value than the natural images and video images etc. However, due to various reasons, in current society,the digital watermarking algorithm about document image is not so much. And the robustness are not good, the capacity of watermark embedding is small.Because of their own characteristics,the document images, and natural images are very different, many watermarking algorithm which apply to the natural image does not apply for the document image, document image recognition includs multiple types of object, the algorithm proposed in this paper is mainly directed against the image formats which Chinese characters are transformed into. Watermarking algorithm proposed in this paper is based on the space domain of the document image, by changing the character of each document in the gray value to achieve the purpose of embedding the watermark, the watermark embedded into the image in this article is a cosine wave. Because of the obvious character features,the stable block features,and the text coherence features of the document image, it has great feasibility to embed watermark by changing the gray value of the characters in the document.. As a result of the algorithm by changing the gray value of each character to achieve the watermark embedding ,the original image must be pre-processed, such as binarization, denoising, tilt correction and segmentation of characters and so on.Therefor, this paper first analysised and studied the basics knowledge of digital watermarking and related theory, introduced the concept of digital watermarking, the basic framework, classification, algorithm, evaluation criteria and attacks to enable readers have a general understanding to the basic knowledge of digital watermarking, in particular, this paper introduced the text watermarking, and then, a detailed analysis of the characteristics of document images and document image pre-processing is given,and then,the watermark embedding algorithm and detection algorithm was presented.We all know,the experiment is the only way to check up if the algorithm is feasible,so in this paper,a lot of experiment was given,and analysis the date,from the experiment result and theoretic analysis ,we can see the algorithm is robust.and do not need the original image in the watermark detection. the algorithm can be used for document image copyright authentication.In other hand, the algorithms in this paper need some improving, for example,the capacity of watermark embedding is not big,it is not a good algorithms for the document image with little characters,For empty image especially,this algorithm does not work.Document image is playing an increasingly important role as a special kind of image vector in current society.Although the digital watermarking in document image is not a progressed technology,but we believe this is a potienal study aspect.With the more and more attention on document image watermarking, document image watermarking will have a bright future.