融合形状与语义相关性的三维模型标注方法

3Dmodel semantic annotation method fusing shape and semantic relevance

为了解决目前三维模型语义标注方法鲁棒性差和精确程度不高的缺点,在三维模型与语义之间相关传递原理的基础上,提出了三维模型与语义词汇之间的相关基数度量算法,并结合骨架结构以及几何细节在模型匹配中的应用,设计了一种三维模型自动语义标注方法。实验验证了该方法具有较高的准确性和鲁棒性,不但在三维模型的形状特征和语义特征之间建立了可靠的映射关系,而且为三维模型的语义检索提供了可靠的依据。

Aiming at the current automatic semantic annotation technology for 3D model being deficiency in robustness and accu- racy, the model-semantic vocabulary related base metrics algorithm is proposed based on the semantic correlation and heir princi- ple among the three-dimensional models, and then a kind of automatic semantic armotation method for 3D model is designed combining skeleton structure with the geometric details in the application of the model matching. Experiments show that this method has high accuracy and robustness, and not only reliable mapping relationship between shape feature and semantic feature of the model is established, but also the reliable basis for the semantic retrieval of 3D models is provided.