Semantic Visualization In3D Urban Environment Research On Labeling Placement

Computer visualization offers us a way to see the unseen things. As computer technology develops,3D visualization results have been wildly used in our daily life such as urban planning, navigation, thematic query and public administration. However, they mainly focus on the representation of the3D environment itself while ignore representation of non-graphic features. Hence the ability for users to acquire information is not improved but impaired, which makes the visualization of non-graphic features, semantic information in3D environment a necessary. In computer visualization field, semantic information is to bridge the gap between the real world and the data, symbols stored on computers. This research chooses text as the representation form of semantic information and deals with3D labeling problems.Currently, the research of3D labeling in China still has a long way to go. Previous researches mainly focused on the automated labeling with digital maps or in virtual environments, which are different from the3D urban environment in this research. Researchers abroad did do some tests on3D labeling while they also just started. Hence there is no standard or criteria for labeling in3D environment. Based on related rules and standards of2D labeling, this research discusses the possibility to place labels in3D environment and applied some of the ideas with OpenSceneGraph. Finally, a way to optimize occlusions among labels is proposed and tested. The contents of this research are:(1) Based on2D labeling rules, this research proposes the needs of3D urban environment labeling and summarized the labeling methods and styles of3D labeling.(2) In accordance with quality assessment and optimization requirements of2D labeling, primary reference for those of3D labeling is proposed.(3) This research analyzes the position selecting process of3D labeling and applied it on screen-based labeling and object-based labeling.(4) For occlusions of labels in3D environment, a filtering algorithm is proposed and applied; the result shows it effectively improves the performance of3D labeling.