中文摘要：

为了解决基于关键字的搜索方式不能准确查找三维模型的问题,对三角网格形式的三维模型提出一种基于内容的模型搜索算法.首先将模型中的三角网格再次划分,增加形状函数的维度,定义了一种新的三维模型形状函数,以提高形状函数的准确性;然后确定三维模型的极轴,将三维模型绕任意轴的旋转转换为绕极轴旋转,增强了球面调和方法提取的形状描述符旋转不变性;最后根据形状描述符之间的欧氏距离来度量三维模型的相似程度.实验结果表明,该算法适用于基于内容的三维模型搜索,并且在准确性和时效性方面都令人满意.

英文摘要：

A content-based algorithm for retrieval of the 3D model composed of triangular meshes is proposed.First we propose a new shape descriptor of the 3D model by remeshing the triangles with increased dimensions.We then determine the polar axis of the 3D model and adopt it to define the shape with spherical harmonic basis functions（SHBF）.It is shown that the coefficients of the SHBF for a particular shape is rotation invariant around the polar axis and can be represented as the shape descriptor.The Euclidean distances of shape descriptors are finally evaluated to compare the similarity between 3D models.Experiments demonstrate that our algorithm provides high accuracy and efficiency for content-based 3D model retrieval.