An automatic classification and recognition method for indoor space
The automatic spatial data extraction methods are missing when indoor maps are created in batches with computer aided design (CAD) building plans as data sources. Therefore, this paper proposed an automatic classification and recognition method for indoor space. Firstly, the essential elements for reconstructing the indoor space layout were automatically screened according to the mapping relationship of the layers, and the obtained door symbols were classified according to the block name and abstracted into adjacent straight lines along the wall, thus restoring wall connectivity. Then, surface elements with functional attributes were constructed by combining essential elements with superimposed annotations. Additionally, five fields were added, including semantics, area, rectangularity, code, and category, so as to record spatial information and functional attribute information of surface elements. Finally, public space, exclusive space, wall space, facility space, and anomalous space were identified by calculating categorial codes. The building plan of a shopping mall was studied, and the experimental results show that this method efficiently processes original drawings, automatically acquires indoor space data, and classifies them based on spatial characteristics.
万方数据
维普
