关键词: 特征提取, 预处理, 类原型, 聚类模型, 三维模型, 聚类算法

Abstract: In view of the practical application problems in three-dimensional model data sets clustering, such as model classification management, the clustering pretreatment for retrieval, etc, a new clustering model based on the classification method and which was featured by unsupervised learning, was proposed. Through adopting feature extraction algorithm grounded on the MFD (Moment Fourier Description) and integrating the existing sophisticated clustering algorithms, this model combined feature extraction with clustering computation effectively, besides taking full considerations of sensitivity issues of clustering data format. The compuatation results indicated that this model could gain local optimality to the certain classified structure data sets on the premise of having overall clustering effect.

Key words: feature extraction, pretreatment, clustering prototype, clustering model, three-dimensional model, clustering algorithm