An effective method for river generalization

Abstract

River generalization is an important part of Geographic Information System. Traditional research is focused on how to extract river systems of electronic maps correctly. They just delete rivers with low levels while not regarding redundant points of rivers of electronic maps or whether the speed of their methods could meet the users' demands. In this paper, we propose a method which filters out redundant information of electronic maps in two steps. First, we make use of a “knowledge decision making” method to extract river systems. Second, we deleted both redundant rivers and redundant points of the maps. In order to speed up our method, we use MPI to design a parallel method to delete redundant points. Experiment results show that our method could extract river systems correctly. With the number of the processes increasing, the parallel based curves rarefy method could increase in speed. Compared to traditional methods, the readability of a map will be improved after filtering out useless information and the storage space that it needs will drop drastically.