Research On Map-Graphics Generalization Based On Raster Mode

After studying the current methods of raster and vector which have been used to map generalization theoretically, it has been found that, the vector methods is difficult to deal with the relationships between features because of the spatial object complexity while structured map generalization methods have been widely recognized. But vector data structure itself is very complex, and it is hard to generalize if all data are not structured which is also a very difficulty job. The structuring matter means that the geographic data organization as well as the required relationships between all data will be structured of which complexity is more than map generalization (Hu Peng,2002). Mathematical morphology among the raster methods is a way to manage the raster with statistical property. It is difficult to execute the exact process if given a determined pixel to which graphics and image will have different results. The majority raster-based methods used in map generalization have their origins in image processing and analysis, for example, the computation of distance and direction neighborhood, overlay and edge detection, the establishment of spline surface and minimum energy surface, low-pass filtering, convolution operator, contraction and expansion. By selecting structural elements, raster methods are in condition of certain constraints (for example, to maintain the shape characteristics) to perform several typical generalization operation, such as delete and smooth objects, conflict detection, aggregation the areal features and displacement.But all these methods have not concerned the complex semantic structure and context of the geographic objects. So, they only have been to deal with simple experimental data. For example, it is difficult to enlarge the partial edge characteristics of areal features based on raster mode. Only approximate generalization result will be gained when the same structure elements act on the polygons which have different shape and size.Therefore, based on the characteristics and advantages about the raster data structure, we have an idea to generalize maps using the methods of combining raster data and progressive cartographic generalization theory. Choose a certain size of structure element as generalizing step, at the same time; build the buffer area of this structure element with an appropriate radius which can form the context environment as to control the local relationships between spatial objects and overcome the short backs of traditional methods. Besides, based on the idea of workflows, the kinds of generalization workflows have been designed according to the different feature classes (semantic) and spatial relationships between them (generalization context) which are used to improve the degree of automation cartography generalization and also means very important and valuable to the production units in practice. Specific research works are as follows:(1) The concepts and aims of cartography generalization are discussed. Characteristics and basic prerequisites of automatic cartography generalization are summarized. Following are state of art of cartographical generalization, and next research directions of automatic cartography generalization.(2) The basic principle and methodology about progressive cartographic generalization based on raster mode are studied. Begin with the spatial abstraction, spatial recognition following the gestalt law is concluded. Following are the basic concept and the methods of control process and the basic algorithm for progressive cartographic generalization. Characteristics of the raster data structure and rasterization algorithms are analyzed. The state of art of cartographical generalization based on raster mode is summarized. A concept framework about progressive cartographic generalization based on raster mode is given after the concept and methods of progressive cartographic generalization based on raster mode have been analyzed.(3)The problems about data mode currently are summarized. The requests for data abstraction and organization of geographical generalization in geographic information system are analyzed. It is concluded that practicable and perfect data mode must be all-important condition of more efficient cartographic generalization process. Two data mode which aim at cartographic generalization are introduced. There isn't any relationship between the multi-scale geospatial databases of our country. Based on the rules of residential areas and roads changing with scales, a new method of matching the residential features and road features are studied to match two features which present the same entity in the two datasets. The methods are implemented to build up the relationships of roads and buildings of the residential features in the NFGIS of the same area.(4)Spatial direction-relationship and spatial distance-relationship based on raster mode are introduced. Topological relations between different types of spatial objects in raster mode are given. The spatial relationship abstraction law of geographical objects based on raster mode is studied. The spatial relationship abstraction law which is produced by cartographic generalization operations is discussed by comparison to vector map.(5)Several methods such as mathematical morphology, map algebra etc which are used in cartographic generalization are analyzed. Kinds of constraint conditions during the process of cartographic generalization are summarized. Following are constraint conditions based on raster mode and the ways to measure. Based on raster mode, the linear object simplification are studied, specially the problem that the linear objects may intersect by itself during the generalization. A method of making use of structure element to recognize is proposed. Take a window of 3*3 size for example, mode to keep line connectivity are summeralized which can preserve local connectivity generalization before and after.Two methods about areal objects simplification and aggregation are introduced.(6)Research on cartographical generalization based on intelligent workflows is studied. Cartographical generalization operations and kinds of methods of cartographical generalization process control are summarized and following are the basic concept of workflow and the applications to geographic information system and geographical generalization. Then, the types of features (semantic) and the spatial relationships between them mainly considering residential area, roads and rivers are anaylized according to the 3*3 structure element. Kinds of cartographical generalization workflows are designed respectively.