CAJ | 학술논문

针对三维模型的分类问题，提出了一种基于核主成分分析（ Kernel-Principal Components A-nalysis，K-PCA）的三维模型分类算法。该算法首先选择形状直径函数（ Shape Diameter Function，SDF）作为特征描述符来提取三维模型的特征向量；然后使用核函数将原始特征向量映射到高维空间中并在该空间上进行PCA得到新的特征向量；最后使用KNN算法并计算未知模型与已知类别的k个模型之间的l2范数以实现模型的分类，确定未知模型的类别。实验结果表明，该算法能够很好的识别三维模型的几何特征，能准确的区分不同类别的三维模型，具有较高的分类准确率。
침대삼유모형적분류문제，제출료일충기우핵주성분분석（ Kernel-Principal Components A-nalysis，K-PCA）적삼유모형분류산법。해산법수선선택형상직경함수（ Shape Diameter Function，SDF）작위특정묘술부래제취삼유모형적특정향량；연후사용핵함수장원시특정향량영사도고유공간중병재해공간상진행PCA득도신적특정향량；최후사용KNN산법병계산미지모형여이지유별적k개모형지간적l2범수이실현모형적분류，학정미지모형적유별。실험결과표명，해산법능구흔호적식별삼유모형적궤하특정，능준학적구분불동유별적삼유모형，구유교고적분류준학솔。
In order to solve the problem of 3D model classification,a new algorithm,which is based on Kernel Prin-cipal Component Analysis( K-PCA),is proposed. The algorithm firstly selects the Shape Diameter Function( SDF) as the feature descriptor to extract the original feature vectors. Then the original feature vectors are mapped to a high-dimensional space,and the PCA is adopted to reduce the dimension of original data for getting the new feature vectors. Finally,the KNN algorithm is applied to find k models in the known 3D model database,and the l2-norm between the unknown 3D model and the k models is calculated to predict the membership of the unknown model. The experimental results show that the algorithm can identify the geometry features of the 3 D models and distin-guish the difference of the models precisely. Besides,the algorithm is highly accurate.