Automatic Clustering of Indoor Area Features in Shopping Malls

The comprehensive expression of indoor maps directly affects the visualization effect map and user’s reading experience. Currently, only points, lines, polygons outdoor are used as objects cartographic generalization. Therefore, this study considers area features generalization deems automatic clustering shopping malls research goal. approach is to construct an encoder-decoder model, where encoder consists a graph convolutional network its variant models. results show that proposed model framework effectively extracts suitable for space improves efficacy. Specifically, with Relational Graph Convolutional Network demonstrated best performance, time complexity, accuracy results, up 95%. This extends object maps, enabling features, proposes important scene malls. valuable scholars interested in maps.