三维模型特征面分级过滤检索技术研究

Research on 3D Model Multi-stage Filtration of Characteristic Face Retrieval Technology

为了弥补基于统计特征提取算法对三维模型的内容描述不够充分,以及识别能力不高的缺陷,提出了三维模型特征面分级过滤检索方法。在线下阶段对IGES格式的模型文件进行解析,得到模型的面属性信息。在线上阶段手绘输入模型特征面对模型进行多级过滤,再根据归类统计的特征面数据表转化的模型特征向量进行精确检索,并通过距离算法进行相似性计算和排序。实验结果表明:该检索方法在效率和准确率上有所提高,并且能够满足一般工程检索的需要。

To compensate the insufficiency of describing the content of 3D models and low recognizing capability based on extraction algorithm of statistical feature, a new approach of 3D model multi-stage filtration of characteristic face retrieval technology is presented. In the off line phase,the model file of IGES format is parsed to get the attribute information of faces. In the on line phase,the characteristic face of model is input with handpainting for multi-stage filtrating. And precise retrieving is performed according to the eigenvector of model transformed from the statistically classified characteristic face data table. The similarity and ranking is calculated with distance algorithm. The experimental results show that the efficiency and accuracy it improved,and can meet the general requirements of engineering retrieval.