The Research Of Registration Of DLG And DOM Based On Linear Feature

DOM , the raster image after geometric correction, has map geometric resolution and rich content; DLG, the vector data file, is widely used because vector data can be used for various spatial analysis. There are several situations where the registration of DLG and DOM is wanted: DLG precision check, a combination of DLG and DOM for field mapping, congruence of DLG and DOM to obtain multiplex relief map.According to the difference of matching primitives, matching approaches are divided into several categories: gray-based matching method, feature-based matching method,etc. The former bases on gray information of local area and the latter is built upon the features extracted from the image. Due to the stability of feature and relative abundance of information, the latter becomes the hotspot of the research. Linear feature can provide more information than dot feature and can be extracted and described conveniently compared with the planar structure, and linear feature plays the leading role in DLG, therefore, a registration method of DLG and DOM based on linear feature is brought forward.In this paper, first the fundamentals of image registration is analyzed, then registration principle and method based on linear feature is discussed, and , the key technologies of registration of DLG and DOM based on linear feature are studied. At first, an improved Canny operator is adopted for edge detection, then two steps : !, rough registration, Hough transformation is used to detect those typical straight lines and then match them, then the endpoints of matched straight lines in DLG and DOM are control points for rough registration. 2, precise registration: after boundary extraction, with certain constraints, feature registration goes on, and the corners and centers of matched boundaries are chosen as the control points for precise registration of DLG and DOM. Finally, by using this approach mentioned in the paper, fine registration results and accurate calibration parameters can be gotten, and the results show that this method is feasible.The purpose of this research is to find a quite good registration method based on linear features, including: a robust line pick-up method, boundary pick-up, boundary representation of chain code, descriptor of shape number, feature matching, registration points pick-up, affine transformation.