Research On Algorithm Of Digital Watermarking In Document Image Based On Segmenting Image And Text

With the increasing popularization of the modern network information technology, digital devices such as scanners and digital cameras become more available, and mass storage media for digital data becomes more affordable, the use of digital images in practical applications is becoming more widespread. Practical imaging applications range from famous works of art, to bank checks, and medical images. Reliable methods for copyright protection, copy control, annotation, and authentication are therefore needed.It became more and more difficult for traditional data encryption method to provide effective protection on the security of multimedia information and data products, and more and more technology of information hiding has been applied. At present, digital watermarking techniques are widely used in authentication, anti-counterfeiting, anti-tamper, protection of data security and integrity as an important branch in the research filed of information hiding and have been applied widely in other fields. It has developed efficiently in academia.However, most of the methods developed today are for grayscale and color images, where the gray level or color value of a selected group of pixels is changed by a small amount without causing visually noticeable artifacts. These techniques cannot be directly applied to binary document images where the pixels have either a 0 or a 1 value. Arbitrarily changing pixels on a binary image causes very noticeable artifacts. Most document images are binary images where the pixels have either a 0 or a 1 value. Until recently, there has been little work on watermarking and data hiding techniques for binary document images. Under the development of information globalization, people need the technique more and more. According to the characteristics of the image, the analysis shows that document images and natural images can be distinguished according to the characteristics in the spatial domain and the frequency domain, and these characteristics are often based on the features of global images. The error when using the local image characteristics is too large, so it is difficult to segment the images mixed graph and text information, but it is particularly important. A grate of document images contains not only text area, but non-text regions as well as pictures, tables and so on. Therefore, it is necessary to segment on document image, adding watermark in text regions and non-text regions.In this paper, firstly, the concept of document image has been introduced, and the classifications of the document image also have been introduced. The present segmentation algorithms of document image have been researched and analyzed. And the concept, characters, classifications, utilized fields, common attacks and typical algorithms of digital watermarking have been introduced in detail. Secondly, the characters of the text and image area of the document image have been analyzed from theirs statistical characteristics, such as the gray histogram, mean, variance, energy, entropy, deflection, peak and so on, and give an arrangement necessary for the embedding after segmentation. Finally, based on the characters of mixed document image, the document image is preprocessed, this is an important step, and it includes: image binarizing with Otsu method, images denosing with median filter, Skew detection with Hough transform, mathematical morphology method is used to find the text region and image region in the document image. Use opening operation to obtain the image region of the document image, and the closing operation to fill the unconnected region in the image region, so as to achieve the segmentation between the image region and the text region. Embed watermark in them separately and test the performance of the algorithm by calculating the Peak Signal to Noise Ratio. At last, we make a series of geometric attacks (noise, filter, cut, rotation) to the document image which has embedded digital watermarking; it proves that the robustness of the digital watermarking is good. According to it, this paper will present a simple, efficient, accurate algorithm.