Applied Research, Document Image Segmentation Based On Graph Theory

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, Document image segmentation research is particularly important.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. That is to say the majority of document images are significantly different by some combination of the characteristics of, Normally contains the text of regional, regional and illustrations chart the region, and the text of regional language also contains the outline of the background color and text, illustrations of the region are usually high-resolution color images,These areas not only in the logical sense of independence, but also in color and spatial resolutions is also up significantly different characteristics. the document image with Obviously image features and independence is more suitable to use characteristics of overall segmentation partitions to deal with.Image segmentation based on graph theory is a newly developed image segmentation technique in recent years. The technology has the characteristics of the overall segmentation. This thesis introduces and illustrates a general framework to integrate the areas of image segmentation and graph-cut theory. The image is mapped into a weighted undirected graph and the pixels are considered as vertexes and the similarity between the visual properties (e.g. gray-level intensity, color or texture) at each pair of neighboring pixels is assigned as the respective edge weight. Therefore the image segmentation can be obtained by cutting the graph with a minimum cut criteria. Specific research on methods of normalized. In order to increase the speed of the interactive segmentation, we introduce a multi-scale method for the computation of graph partition that is motivated by the well-known multi-resolution signal processing theory and the pyramid structure. We perform a number of numerical experiments to show that this multi-scale computation method can reduce the running time of the segmentation algorithm, as well as can produce nearly the same segmentation result as the conventional graph cuts method, especially when the object to be extracted is small compared to the whole image.