轴类零件图特征的识别与匹配  

Feature Recognition and Matching Method of Shaft Parts Drawing

作　　者：周超 许黎明[1] 乔文俊 时轮[1] 王北辰 ZHOU Chao;XU Liming;QIAO Wenjun;SHI Lun;WANG Beichen(School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China;Shanghai SmartState Technology Co.,Ltd.,Shanghai 201306,China)

机构地区：[1]上海交通大学机械与动力工程学院,上海200240 [2]上海交大智邦科技有限公司,上海201306

出　　处：《机械设计与研究》2022年第5期126-129,共4页Machine Design And Research

基　　金：上海市人工智能创新发展专项(2019-RGZN-01026)。

摘　　要：现有的二维工程图纸研究多集中于图纸分类与检索、三维模型重建、图纸符号定位与识别等,图纸中具体的语义特征具有复杂多变的特性,致使难以建立一套规则对尺寸标注信息与特征轮廓进行匹配。为实现光栅格式轴类图纸语义特征的自动读取,提出了一种基于轮廊检测的轴类零件图特征识别与匹配方法。通过对图纸进行二值化、开闭运算、Canny轮廊检测、霍夫圆检测等操作,提取并识别图纸中的矩形轮廊和圆弧轮廓;通过分析各个轮廊与周围尺寸标注间的大小与相对位置关系完成匹配。实验结果表明,该方法对符合制图规范的阶梯实心轴图纸的轮廊平均匹配准确率达到89.8%左右。The existing research on two-dimensional engineering drawings mostly focuses on drawing classification and retrieval,three-dimensional model reconstruction,drawing symbol positioning and recognition,etc.The complex and changeable characteristics of the specific semantic features in the drawings make it difficult to establish a set of rules to match the dimension information with the feature contour.In order to realize the automatic reading of semantic features of shaft drawings in raster format,a feature recognition and matching method of shaft parts drawings based on contour detection is proposed in this paper.The rectangular and arc contours in the drawing are extracted and recognized through binarization,opening and closing operation,canny contour detection,Hough circle detection and other operations on the drawing.The matching is completed by analyzing the size and relative position relation between each contour and the surrounding dimension marks.The experimental results show that the average matching accuracy is about 89.8%for the stepped solid axis drawing contour in accordance with the drawing specification.

关 键 词：图纸识别 特征匹配 特征提取 图像处理 

分 类 号：TH126.1[机械工程—机械设计及理论]