Research On Terrain Sketch Visualization And Generalization Based On Visual Depth

Terrain sketch is rendered in the pictorial way for the natural terrain surface and its morphologic features. Respecting its simplicity, perception, abstraction and artistry, terrain sketch is widely used by geographers to depict the natural terrain surface. Even nowadays with advanced photographing technology, it is still in popular use. Traditionally, it was almost painted manually while new methods based on photographs or Digital Elevation Model (DEM) for automatic terrain sketch rendering are presented. However, as a primary issue of vision, visual depth based generalization on terrain sketch is usually ignored in many studies.As a visual abstraction on alternative terrain feature characteristics, terrain sketch is also delivering a natural vision law that being further away from an object results in less detailed information being observed, namely visual depth based generalization. Visual depth determines how many details of terrain sketch are presented. Hence, this study is mainly focused on the definition of digital terrain sketch, realization of automatic rendering and visual depth based generalization on terrain sketch. In order to achieve these objectives, this study firstly performs analysis on the morphologic and visual characteristics of terrain sketch. Then, terrain sketch features are defined and their extraction algorithms based on DEM are followed. The primary visual factors of generalization process on terrain sketch are presented after simulating and analyzing the natural visual generalization of our eyes. In the perspective of the formation principle of terrain sketch, this study presents two generalization algorithms with a function of visual factors. One is for the terrain sketch features and another is for the terrain surface. Both algorithms implement generalization process on the feature datasets. An automatic rendering framework is developed for performing the experiment works and for evaluating the efficiency of terrain sketch models and generalization methods.The main works in this study are listed as follow:1. Extraction algorithms for terrain skeleton lines and visual occlusion contours are presented and performed. Based on the analysis of terrain morphologic and visual characteristics, the terrain skeleton lines are regarded as one component of terrain sketch features and extracted using digital terrain analysis methods. The visual occlusion contours are extracted as a line structure by adopting a substitution method.2. Visual depth based generalization on terrain sketch makes a difference. Through an analysis on the natural visual generalization, the primary visual factors are defined as the dominating parameters in generalization algorithms. These presented algorithms in this study perform expectedly on generalizing terrain sketch.3. A utility framework is developed for automatically rendering terrain sketch scenes, shading scenes and point cloud scenes. Using C++ programming language and graphic library OpenGL, this study develops this framework to extract sketch features, render terrain scenes and perform visual depth based generalization on terrain sketch. For the only sake of experiments, the execution algorithms in this framework are not optimized.