Abstract：The acquisition and expression of local region features is very important for the study of 3D CAD model clustering.To solve the problem of local region feature representation，this paper extend the existing six-tuple method to seven-tuple based on the existing six-tuple method，and add the edge attribute information formed by the intersection of the surface and the surface in the model，so as to obtain the vocabulary book constructed by the local region feature better.In the clustering stage，this paper propose a model local region weighting method，which reduces the common local regions in the clustering phase.Similarity calculation is the most important degree，which relatively improves the more differentiated local areas.The experimental results show that the proposed method can effectively support the clustering task of CAD model.Compared with the baseline method，the NMI，V-measure and Purity values of four typical clustering algorithms are improved.
汪大涵,王裴岩,张桂平,马伟芳. 融合权重信息的三维CAD模型聚类研究[J]. 小型微型计算机系统, 2020, 41(6): 1296-1301.
WANG Da-han,WANG Pei-yan,ZHANG Gui-ping,MA Wei-fang. Research on 3D CAD Model Clustering Based on Weight Information. Journal of Chinese Computer Systems, 2020, 41(6): 1296-1301.
2020, Vol. 41 
