Application Of Chain Code Technique And Region-labeling-automaton To Image Processing

With the continual advances and maturation in digital image processing and pattern recognition, there is a speedy advance in image coding technique which is utilized more and more in the field of digital image processing by its fine characteristic .After an analysis of the theory and advances of digital image processing, the thesis introduces in detail how a two-value image is expressed to the chain code form and the characteristic of the method. On the basis of region-labeling-automata invented by Gu Guo-qing and Xu Yan-bing, applying the vertex chain code technique, an algorithm for detecting and locating corners and a set of new stitching algorithm for document images are proposed.Applying the characteristic of vertex chain code of region borderline, the algorithm for detecting corners can accurately detect the corners of the regions which borderlines are all beelines in two-value images .The stitching algorithms for images expatiate the stitching theory and technique for gray or two-value document images .The two-value document images include tilt and without tilt images. The stitching algorithms can fix the matching points of adjoining overlapping images using the method of vertex chain code and region-labeling-automata cooperating with maths or image processing technique. In the research of stitching for two-value images a method of circumgyrating for two-value images is given, which applying the characteristic of corners of regions.Combining the above algorithms technique and theory, adopting the programme method of Visual C++,we develop a experiment system for above algorithms which function include detecting corners of region of two-value images, stitching gray and two-value document images and circumgyrating two-value images in which there only are region which borderlines are all beelines.