基于PHOG特征的电路图中电气符号识别  

Electrical Symbol Recognition in Circuit Diagram Based on PHOG Feature

作　　者：肖豆[1] 侯晓荣[1] 

机构地区：[1]电子科技大学能源科学与工程学院,成都610054

出　　处：《舰船电子工程》2017年第1期90-93,共4页Ship Electronic Engineering

摘　　要：针对电气符号的大小、图纸背景的模糊、电气符号的旋转角度等各种干扰因素对计算机识别电气图纸造成的误差问题,提出一种基于提取塔式梯度方向直方图(Pyramid Histogram of Oriented Gradients,PHOG)特征的电气符号识别方法。首先运用直方图分析和形态学处理的方法,分割出电路图中的电气符号。其次建立电气符号训练集,提取电气符号图像的PHOG特征。最后使用这些PHOG特征和分类信息对支持向量机进行训练,利用支持向量机进行识别。结果显示PHOG算法对电气符号具有较高的识别率,与已有的一些方法比较,识别效果更好。According to the size of the symbol,electrical drawings,electrical background fuzzy symbols such as rotation of various interference factors on the computer recognition error caused by electrical drawings,a histogram extraction tower is presented based on gradient direction（Pyramid Histogram of Oriented Gradients,PHOG）electrical characteristics of symbol recognition method.Firstly,the method of histogram analysis and morphological processing is used to segment the electrical symbols in the circuit diagram.Secondly,it sets up the electrical symbol training set,and then extracts the PHOG features of the electrical symbol image.Finally,these PHOG features and classification information are used on the support vector machine training,it uses support vector machines to identify these symbols.The results show that the PHOG algorithm has a high recognition rate for electrical symbols,and the recognition effect is better than the existing methods.

关 键 词：电气符号 形态学 PHOG特征 支持向量机 

分 类 号：TN60[电子电信—电路与系统]