| Large-scale scene modeling is currently a hot issue in computer research, but also a complex issue related to many disciplines. To restore the real scene of architecture by virtual simulation is of big significance to advertise, protect and inherit the traditional ancient architecture.However, in existed modeling concepts, no applicable modeling method is on large-scale scene of ancient architecture automatic modeling.Basing on the analysis and researching of Anhui-Styled architecture, this thesis summarized and analyzed various modular and parameterized modeling methods in recent, proposed an automation modeling method for characteristic buildings which based on subgraph expansion.First, taking the user simply inputs as initial topological information, topological graph is automatically generated through association rules by subgraph expansion method, by the way module attributes of each node reasoned out by the various types of constraint rules, then traverse the final topology and generated scene description file, at the last rendering the scene and getting Anhui-Styled architecture model, what effectively improve the modeling accuracy and speed.This algorithm obviously improved the automatic modeling structural unity problem, and also effectively improved the modeling accuracy and speed.Association rules are used to describe the relations among the things in data sets, also the basis of sub-graph expansion method. By optimizing linking items of the frequent item set, and compressing the things item sets improved Apriori mining algorithm,and an experimental result were given to approve the effect. At last, combining the data structure characteristic of Anhui-Styled architecture topology, a method that combination of triples and hash tables to improve the storage structure of the graph is proposed,which had throuthed the analysis experiment that proved this storage method can achieve better storage efficiency. |
