半监督的三维网格模型层次分割.pdfVIP

22 4 Vol. 22 No. 4 2010 4 Journal of ComputerAided Design Computer Graphics Apr. 2010 1, 2) 2) 1, 3) 孙晓鹏 张 琪 魏小鹏 , , 1) ( 116024) 2) ( 11602 ) 3) ( 116622) ( xpzhsun@ gmail. com) : K , 3 . ; K , ; , , . , , . : ; ; ; K ; : T P3 1 Semisupervised 3D Mesh Hierarchical Segmentation 1, 2) 2) 1, 3) Sun Xiaopeng , Zhang Qi , and Wei Xiaopeng 1) (School of Mechanical E ngineer ing, Dalian University of Technology , Dalian 116024) 2) (Dep artment of Comp uter and I nf ormation Technology , iaoning Normal University , Dalian 11602 ) 3) (K ey aboratory of Advanced Design and I ntelligent Comp uting (Dalian University ) , M inistry of Education, Dalian 116622) Abstract: T his paper presents a novel and efficient algorithm for the hierarchical mesh segmentation of 3D meshes, w hich is based on semisupervised Kmeans clustering and kring strip grow ing . T he new technique consists of three consecutive stages: prominent feature point extraction, presegmentation and postsegmentation. T he prominent feature points are extracted using multidimensional scaling transformation; the original 3D mesh is then segmented initially using the semisupervised Kmeans clustering algorithm in the presegmentation stage to improve the algorithms efficiency. According to the presegmentation results, w e apply the Gaussian curvature and kring strip grow ing algorithms to get a hierarchical segmentation of the mesh. Comparing to the other algorithms, our proposed algorithm can obtain better boundaries, and the regions are harmonic. Key words: h