Research On Linear Feature Continuous Scale Transformation By Preserving Spatial Context

The rise of portable media devices poses a higher demand for interactive maps.Media devices evolve from the traditional PC to the digital screen,smart TV,high-resolution PC,smart phones,smart watches,etc.We have got different size of screen and screen resolution also differs,which result in various interactive needs.Traditional map carrier is mainly paper and PC,so,the map data processing mainly based on a few fixed scale,which cannot meet the current map interactive needs.People want to zoom in and out while keeping the map focus,so the continuous map generalization technique is presented to provide arbitrary levels of scale map data.Map generalization belongs to the scale transformation.At the same time,map is an abstract expression of the real world,relying on the planet earth,any geographical entity or phenomenon exists in a specific geographical environment,we call it context.So,scale transformation must also take into account its geographical context and its own characteristics.Therefore,this paper studies the continuous scale transformation of map based on the context of geographical features.The main research contents are as follows:1)Summarizes two ways to continuous scale transformation.First,change the traditional data storage approach to explore the hierarchical data structure to support continuous resolution map output;Second,based on the multi-version of the data to establish the relevant model and algorithm to export arbitrary scale map data.At present,the multi-version data is based on the idea of Multi-scale fusion in the field of remote sensing and image processing.The multi-version data interpolation is used to generate intermediate scale map data to achieve continuous scale transformation.Morphing is applied as a technical means.Morphing consists of two steps,matching and interpolation.Current matching algorithms either simply consider the feature's geometry,or the matching process is complex.There is a need for a matching method that takes the spatial context of the feature into account.2)Presents a detailed illustration of scale and context-related theory.Scale is the most important factor that affecting the expression of the map,we analyze the connotation and extensions of scale.The impacts on spatial cognitive and multi-scale expression are also considered.Four kinds of scale transformation modes are summarized.Finally,concludes the system framework of the spatial context from three different aspects:micro,medium and macro.3)Proposes a morphing method that takes into account the overall context of the map feature and is used for continuous scale transformation of linear features.Morphing consists two main steps:matching and interpolation,and matching results will affect morphing's final outcome.Puts forward a new matching method which takes into account the overall context of the map feature,uses the shape context as the shape descriptor,gets the matching points through the histogram matching.The method does not need special points,which is more robust.Experiments show that the method can improve the matching and then improve the morphing accuracy.4)Introduces a shape matching method by preserving local neighborhood structures,and apply it to the morphing of polylines.The map from the graphical point of view is a graphical affine transformation process.During map generalization,although the absolute distance between the two points will change a lot,but the point's neighborhood structure and context information are stable.Considered the Tobler's first law of geography,we need to keep the neighborhood structure information in the process of generalization.Using shape context descriptor to get the matching costs between points which then can be used to initialize the matching probabilities matrix during relaxation labelling.Afterwards,iterate the support functions to update the matching probabilities matrix until we get the optimal matching results.The two polylines are divided into two groups of sub-segments by the matching results.Finally,using the linear interpolation method to morphing every pair of the corresponding sub-segments.Extensive experiments show our method can well preserve the global context and local neighborhood structures,improving the accuracy of morphing transformation.