Research On Object Simplification And Scale Sampling Method In LOD Model For Vector Data

Nowadays, the rapid developing of "Digital Earth" and Internet technique has made the conflict between the limited national information infrastructures and massive spatial data more sharp. Building LOD(levels of detail) model of GIS data with LOD technology is a crucial way to solve the problem since that it can realize the compressing and recurring of spatial data. It could help to improve the speed of inquiring and displaying of GIS data. But there are two common problems in LOD model needed to be solved: topological consistency cannot be maintained when simplifying objects and insensible Scale Sampling method. And the two have great effect on the development of LOD model and Cartography Generalization. This thesis has deep research on the two problems. There are already some experts come up some solutions for simplification of object, but most of which are not mature enough and need to be further studied. In addition, most of the Scale Sampling methods presented at present do not consider the characteristics of spatial data, and exist big data redundancy.Aiming at the shortage of the existed simplification methods, this thesis puts forward a method based on M value. The classical Douglas-Peucker was chose to be the simplification algorithm of the method, as it is efficient and keeps the inherent feather of vector graphics well. But when it deals with polygon data stored in PostGIS/PostgreSQL, graphics distortion phenomenon appears at the mutual border between polygons on occasion. This problem was solved with this method and the speed of deriving spatial data was improved through using the Douglas-Peucker to simplify the mutual border and recording the level information of points of Geometry. As to the Scale Sampling problem, this thesis puts forward a method called Quantitative method which takes full account of the characteristics of spatial data. This method uses Spatial Data Amount (SDA) to control the Runtime to get scale value of every level. This method decrease data redundancy and the information capacity in displaying window. And it also improves the transmission efficiency and meets user's need. At last, the thesis builds a prototype system to testify the two methods with experiment, the result shows both of them have solved the problems well.