Aiming at the problem of attending to matching precision,retrieval speed and algorithm complexity simultaneously in the most of current 3D retrieval algorithms,a 3D model multi-feature matching algorithm based on orthogonal views is proposed. First,the front views are obtained by projecting six positive angles of the retrieved model. Second,the gray image is used to illustrate every front view’ s features and the gaps between facets and edges after the projection are filled according to the distribution of the model’ s facets. Then the features of every gray image’ s projection histograms and Zernike moments are extracted dynamically. Finally,based on the analysis of ad-vantages and disadvantages of each feature,multiple features are combined to match the model’ s similarity. The experiment comparison is made with the LFD algorithm and D2 algorithm. The results show that the algorithm suggested can better improve the recall ratio and pre-cision ratio of the model’ s retrieval and it has the advantages of low computational complexity,high matching precision and rapid compu-ting speed.%针对现今多数三维检索算法在匹配精度、检索速度以及算法复杂度三者难以相互兼顾的问题，文中提出一种基于正交视图的三维模型多特征匹配算法。首先对被检索三维模型的6个正视角进行投影，获得6个正视图像；接着分别用灰度级图像来描述各个正视图特征，并根据投影后模型各面的分布情况填补顶点与面片之间的间隙；然后动态提取每个灰度级图像的投影直方图特征和Zernike矩特征；最后在分析各特征的优缺点的基础上，融合多个特征来匹配三维模型的相似度。与传统LFD算法和D2算法进行对比实验，结果表明，文中算法能较好地提高对三维模型检索的查全率与查准率，具有计算量小、匹配精度高、运行速度快的优点。
展开▼
