Inproceedings,
Link based conflation of geographic datasets
Proceedings of the 8th ICA Workshop on Generalisation and Multiple Representation, July 7--8, 2005, A Coruña, Spain, (2005)
Abstract
The conflation of geographic data from different sources is an important research area in geographic information sciences. In recent time this is increasingly motivated by the demand for a higher usability of expensively collected data on the part of the national mapping agencies. For this reason also the linkage of geographic datasets with different scales and their mutual administration in one database is under examination. Linking objects of different scales will ease the propagation of new information to a generalized dataset and the transfer of information from lower to higher scale becomes more feasible. In this paper a method for a topological consistent transfer of features is presented, which assumes, that links between corresponding features exist. The method is based on a map transformation which considers locally existing geometric differences between the two involved datasets. Even though algorithms and mathematical formulas of similar existing methods differ among each other and in relation to the presented work, the term “rubber sheeting?? is often used for this kind of transformation and will also be used here. It refers to the illustration of the map as a flexible membrane that is fitted into a frame and hence is forced to become deformed. Rubber sheeting can be implemented by an interpolation of coordinate differences that are given for control points. In this work an infinite number of control points on corresponding linear elements is defined. After an overview on related work, this paper addresses how the transformation can be set up by exploiting links between linear features. Also the mathematical foundations for the transformation are explained and results for existing datasets are shown. In the conclusion limits are presented that require further development and research.
