基于改进连通域检测算法的电力工程图纸识别技术  

Technology based on improved connected component labeling algorithm for power engineering drawings recognition

作　　者：樊冬梅[1,2] 严文豆 梁毅 刘巍巍 FAN Dongmei;YAN Wendou;LIANG Yi;LIU Weiwei(College of Electrical Engineering,Zhejiang University,Hangzhou 310027,Zhejiang,China;Guangzhou Power Supply Bureau,Guangdong Power Grid Co.,Ltd.,Guangzhou 510180,Guangdong,China)

机构地区：[1]浙江大学电气工程学院,浙江杭州310027 [2]广东电网有限责任公司广州供电局,广东广州510180

出　　处：《沈阳工业大学学报》2024年第4期361-366,共6页Journal of Shenyang University of Technology

基　　金：广东省科技计划项目(GZHKJXM20210046)。

摘　　要：针对电力工程图纸识别中存在无法整体标记粘连字符以及标记效率较低的问题,提出了基于数学形态膨胀处理及简化决策树的连通域标记算法。利用可灵活设定的结构元素对输入图像进行膨胀处理,实现了对粘连字符的整体标记;通过分析相邻像素的各种可能性,设计了一种基于简化决策树的高效连通域标记算法,进而提升了电力工程图纸的识别效率。仿真结果表明,与传统方法相比,所提算法在实现整体标记粘连字符的同时,显著提高了算法的执行效率,其执行时间约为传统方法的1/2。To deal with the problems of overall labeling of connected characters and low labeling efficiency in the recognition of power engineering drawings,a connected component labeling algorithm based on mathematical morphological dilation and simplified decision tree was proposed in the work.The structural elements,which could be flexibly set up,were employed to dilate the input image and further to realize the overall labeling of the connected characters.Through the analysis of various possibilities of the adjacent pixels,an efficient connected component labeling algorithm based on simplified decision tree was designed,thus improving the recognition efficiency of power engineering drawings.The simulation results show that the as-proposed algorithm can significantly improve the execution efficiency when achieving the overall labeling of connected characters in comparison with the traditional methods,and its execution time is about half that of the traditional methods.

关 键 词：电力工程图纸 字符定位 字符提取 连通域标记 粘连字符 决策树简化 数学形态膨胀 标记效率 

分 类 号：TM769[电气工程—电力系统及自动化]