高级检索
孙聪, 施侃乐, 雍俊海. 基于双层结构的矢量工程图符号识别算法[J]. 计算机辅助设计与图形学学报, 2017, 29(12): 2171-2179. DOI: 10.3724/SP.J.1089.2017.16550
引用本文: 孙聪, 施侃乐, 雍俊海. 基于双层结构的矢量工程图符号识别算法[J]. 计算机辅助设计与图形学学报, 2017, 29(12): 2171-2179. DOI: 10.3724/SP.J.1089.2017.16550
Sun Cong, Shi Kanle, Yong Junhai. Algorithm for Recognizing Symbols from Vector Engineering Drawings Based on a Two-Layer Structure[J]. Journal of Computer-Aided Design & Computer Graphics, 2017, 29(12): 2171-2179. DOI: 10.3724/SP.J.1089.2017.16550
Citation: Sun Cong, Shi Kanle, Yong Junhai. Algorithm for Recognizing Symbols from Vector Engineering Drawings Based on a Two-Layer Structure[J]. Journal of Computer-Aided Design & Computer Graphics, 2017, 29(12): 2171-2179. DOI: 10.3724/SP.J.1089.2017.16550

基于双层结构的矢量工程图符号识别算法

Algorithm for Recognizing Symbols from Vector Engineering Drawings Based on a Two-Layer Structure

  • 摘要: 为了从矢量工程图中自动提取零件种类和个数等信息,提出一种矢量工程图符号识别算法.首先定义矢量工程图中每个元素的3种共同抽象属性以及2个元素之间的关系描述子;然后提出描述符号的双层结构,其比传统的网状或树状结构简单;最后,通过关系描述子的匹配与双层结构的构造进行符号识别.另外,该算法利用四叉树、一次识别多种符号以及对关系描述子进行排序的方法加速符号识别.实验结果表明,文中算法比已有算法识别的符号种类更多,识别效率更高,而且在旋转和缩放变换条件下更加鲁棒.

     

    Abstract: In order to automatically obtain types and numbers of parts in vector engineering drawings, we propose an algorithm for recognizing symbols. At first, we define three common properties for each element and a relation descriptor between two elements. A two-layer structure is introduced to simplify the representation of symbols, which are traditionally represented by constraint-networks or trees. Then, symbols are recognized via matching relation descriptors and building two-layer structures of symbols. We also use the structure of quad-trees, a method of recognizing several types of symbols in one cycle, and a method of sorting relation descriptors to speed up symbol recognition. Experimental results illustrate that the proposed algorithm can recognize more types of symbols, requires less time, and is more robust under rotation and scaling transforms than the previous algorithms.

     

/

返回文章
返回