| Abstract： |
Point feature selection aims at preserving global patterns of point cluster during map scaling. This is an important technique for clear presentation of points in multi-scale maps. However, existing methods tend to include single point features ignoring the spatial interactions between two types of points. For example, based on the Input Sharing mechanism in agglomeration economics, different types of facilities are usually co-located together to reinforce their functions in business competition. In this respect, generalization of a point feature should consider not only its own importance but also the reinforcing effects from other features. In this paper, an improved method is proposed by including the spatial contextual information. Experiments suggest that our method can preserve service patterns after generalization. |
