On Electronic Sand Table System From Military Topographic Paper Maps

The digital image processing technique has been widely applied in many fields.The automatic recognition of military maps plays a great role in the construction of the national defense, helping transfer pictorial maps to digital maps and provide data for the building of three-dimensional maps. The automatic recognition of maps is difficult but important to deal with, involving many disciplines and concerning both theories and practice. The automatic recognition of military maps requires high intelligence, better generality, less human involvement and rapid construction in the whole processin particular. This dissertation, based on fuzzy clustering and mathematical morphology, etc. deals with the digital image processing technique in map recognition. Map segmentation algorithm and thinning algorithm etc. are studied in the meanwhile, and, by combining the above methods, the electronic sand table system from military topographic paper maps is developed, which is a good examination of the algorithm involved in the dissertation as well as a good foundation for the further development and improvement of the system.In addressing the problem of correcting the nonlinear geometric distortions of the military topography maps, this dissertation uses a flexible method of the spatial transformation corresponding to the’quadrilateral regions’ tie points. The geometric distortion process over the entire image plane generally cannot describe in single analytical function. The method used frequently to overcome this difficulty, then, used bilinear gray-level assignment, deals with the assignment of gray levels to pixels in the spatially transformed image. The specified location of the corresponding tie points, which are a subset of pixels whose location in the distorted and corrected images is generally diverse and complex. This thesis extracts and calculates the notes of grid, which is a unique feature in the military topography maps to solve this problem. In order to correct the distorted rotation angle, this part creatively uses the image recognition of the digital tagging of the latitude and of the longitude in the military topography maps.In addressing the problem of adjusting the matching spatial pixels between two images, this dissertation tests and evaluates the similarities of standard scopes between images stitched according to the matching principles. How to choose the scope including the control point is a difficult problem. In order to overcome the difficulty, this thesis proposes an effective method to automatically find the sub-region which contains the greatest pieces of color information. Based on the matching technology by correlation, this design uses some synthetic types of technology such as ImageStone, BOOST, MFC programming and so on, and develops an visual C++6.0integrated environment in windows XP operating system, implementing an image stitching software system of military topography map which contains three main function-automatic stitching, semi-automatic stitching, and manual stitching.During the research of color segmentation algorithm, the color space of the map image is transformed firstly. Then, an improved fuzzy c-means (FCM) algorithm based on weighing of swatch is presented. The subordinate degree is adjusted and the2-D color histogram is introduced into the weight coefficient. Two shortcomings in FCM are overcome. That is, segmentation becomes fuzzy when the subordinate degrees become similar and for different sample vectors the clustering effects are different. The proposed algorithm is programmed on computer with Visual C++6.0and a good result is obtained.During the research of thinning algorithm, an improved fast parallel thinning algorithm is proposed. The symmetry of thinning model is taken into consideration. The deleting array is given and the operation speed of the algorithm is improved. The algorithm is programmed with Visual C++6.0with an ideal result. There is neither distorted skeleton nor excessive corrosion phenomenon, and the connectedness is satisfied.In the research of vectorization algorithm, a variable step-size adaptive vectorization based on circle tracing is presented. The size of circle radius is adjusted in accordance with the complexity of curves. Larger circle radius is chosen when the slope of contour becomes smaller. Smaller circle radius is chosen when the slope of contour becomes larger. The representative data is gathered and the character of contour is assured by variable step-size method.This dissertation introduces a method of generating3D images by using contour. It presents the conception and one of the generation algorithms of Delaunay triangulation, and rebuilds the digital terrain model by using interpolation algorithm, which makes use of Open GL and Visual C++software to carry out modeling of one military map and producing a three-dimensional electronic map.Finally, the research work undertaken to complete this dissertation is concluded; the present status and the development of the map digitalization are elaborated.