中文摘要：

为了提高和改进三维模型自动语义标注技术,弥补已有的模型自动语义标注方法在精确性和鲁棒性方面的不足,在三维模型与语义词汇相关传递原理的基础上,提出一种融合相似词汇概率、模型语义直接相关性和模型整体相似性的三维模型与语义词汇之间的相似度计算算法。并结合模型骨架结构的拓扑属性与局部突起的几何形状属性在模型检索中的应用,设计出一种新的三维模型自动语义标注方法。一系列的实验结果验证了该方法的准确性与鲁棒性。

英文摘要：

In order to improve automatic semantic annotation technology of 3D model, and to make up for the deficiencies in robustness and accuracy of existing semantic annotation method of the model, we propose an algorithm based on the principle of correlation transitivity between three-dimensional model and semantic words for calculating the similarity between 3D model and semantic words, which mixes the probability of similar words, the models-semantics direct correlation and models overall similarity. Besides, we design a new automatic 3D model semantic annotation method in combination with the application of topological property of model＇ s skeleton structure and partial protruding geometrical shape property in model retrieval. A series of experimental results verify the accuracy and robustness of the method.