融合Fuzzy拓扑与GALIF的三维形状检索

THREE-DIMENSIONAL SHAPE RETRIEVAL BASED ON FUZZY TOPOLOGY AND GALIF

针对手绘草图检索三维模型时存在的表达模糊性和绘制随意性等问题,提出一种Fuzzy拓扑和局部几何特征相结合的检索方法。首先,通过提取二元图元的Fuzzy拓扑关系获得模型的整体拓扑特征描述,并优化二元图元特征向量的冗余问题;然后,在GALIF基础上,融合角点信息提出CGALIF局部描述子,进一步提高检索的识别率和鲁棒性;最后,针对模型多视角渲染的独特性,提出节点近邻遍历算法。结果表明,该方法具备较高的检索性能。

Retrieving three-dimensional models according to freehand sketch has the problems such as fuzziness in expression and arbitrari- ness in drawing. In order to solve the problem,in this paper we proposed a retrieval method which combines Fuzzy topology and local geometry feature. First ,we obtained the entire topological characteristics description of the model by extracting Fuzzy topological relations of binary prim- itive, and optimised the redundancy problem of binary primitive＇ s feature vectors. Then, on the basis of GALIF we put forward local descriptor of CGALIF by fusing angular point information to further improve the reeoguition rate and robustness of retrieval. Finally, we proposed the neighbour node traversal algorithm according to the uniqueness of model＇ s multi-perspective rendering. Results showed that this method has higher recognition performance.