Layer Segmentation Of Document Images

With the development of information technology, document images are widely used in the projects of OA (Office Automation), DA (Digital Library), EC (Electronic Commerce), electronic government, etc. In these applications, documents of paper are usually scanned into digital document images to be stored, transmitted, displayed and printed. In order to ensure the effectively process of the document image, the image should be compressed a great deal under the premise that which can be identified.The thesis analysis the character of document images, find that the images are made of some regions with special characters. The regions include text blocks, line graph blocks, continue-tone images blocks (true color image) and half-tone images blocks (images with color palette). Text and line graphs blocks reserve more details and structure information, which would demand more space resolution, but little color resolution. Comparatively, the regions of continue-tone images blocks or half-tone images blocks are with more color information, usually demand more color resolution, but little space resolution. Base on the fact, a new compression method is proposed. The new means emphasize on the segmentation of the document images, which is the focus of this thesis.According to the propositional stands for layer segmentation brought forward by the ITU (International Telecommunications Union), in 1997, and composing the arithmetic of document images segmentation presented at the present time, an optimized images of document segmentation model is proposed in this thesis. In this model, the block segmentation method and the layer segmentation method are combined, and an improved multi-scale two color clustering method is used to segment the text region separated from the color images by the block segmentation method. Through the process, it can get outline of text area and background color. Then logic consolidation will be used on the background and graph region divided by the block segmentation. The optimized method combined characters of block segmentation and layer segmentation, could get over disadvantage of present document images segmentation method and get better layer segmentation effect, which could give good base for the continue process. Base on the optimized method of document images' segmentation, the thesis researched the multi-scale thought in the layer segmentation. The images and image features are represented by the structure of tree. The more accurate layer segmentation result of document image is the use of the Bayes images segmentation method based on the Markov random field.At the end of the thesis, segmentation experiments for the optimized model completed and gave the example of the segmentation. Compare with the only block segmentation and layer segmentation, it would be found that the model in this thesis would be more effective.