针对现有检索算法没有充分考虑模型局部信息的问题，文中提出一种改进的特征点分割方法进行特征生成树构造，并用于三维模型检索.首先提取模型特征点和核心部分，再计算每个三角片的平坦度，以特征面片为种子面片，平坦度差值作为增长因子，使用分水岭方法分割模型，可以得到模型各个部分的曲面集合，之后利用集合中各曲面间的拓扑关系创建特征树，最终去比较不同模型的特征树得到它们的匹配度，根据匹配度进行三维模型检索.文中方法在 Visual C++6.0环境下实现.实验结果表明，文中方法有效利用了模型的局部信息，在相同查全率下有较高的查准率，得到较好的结果.%Aiming at 3D model didn't fully consider the model's partial information for the current retrieval algorithms,present an im-proved method of feature point segmentation to construct mesh map for 3D model retrieval. First feature points and the core part of the model are extracted,and then compute each triangular piece flatness,regarding feature point as seed patch,flatness difference value as a growth factor,using watershed segmentation model method,can get each part of the collection of surface for the model,and then use the set of the topological relationship between surface to create characteristic tree,and ultimately compare feature tree of different models to get their matching degree,according to which is retrieved. The algorithm is implemented in the developing environment of Visual C++ 6. 0. Experiments show that the method can effectively utilize the partial information of model,obtain more precision on the same recall and better result.
展开▼
