一种鲁棒的线纹尺图像刻线提取算法研究  

A Robust Approach for Line Scale Image Extracting

作　　者：张晓磊 孙鹏 孙镱玮 邹志[2] ZHANG Xiaolei;SUN Peng;SUN Yiwei;ZOU Zhi(CRRC Qingdao Sifang CO.,LTD,Metrological Physical and Chemical Testing Center,Qingdao 266109,China;Changcheng Institute of Metrology&Measurement,Beijing 100095,China)

机构地区：[1]中车青岛四方机车车辆股份有限公司计量理化检测中心,山东青岛266109 [2]航空工业北京长城计量测试技术研究所,北京100095

出　　处：《计测技术》2020年第1期42-45,共4页Metrology & Measurement Technology

摘　　要：为解决光照不均匀、锈斑、污损等因素造成的钢卷尺图像检测识别不准确问题,提出了一种基于Cabor变换的钢刻线提取方法。通过Cabor变换,使特征方向与特定频率的线条结构被过滤呈现出来,基于这些线条结构计算图像中的线纹尺待检刻线位置,之后通过RANSAC算法拟合刻线的双边缘,从而准确获得刻线中心位置。经实验证明,该算法可以在刻线质量较差的情况下,鲁棒地识别破损和有锈斑的刻线,有效地提高检测准确度和整个测量系统的鲁棒性,具有技术参考价值。In order to solve the problem of inaccurate detection and recognition of steel tape image caused by uneven illumination,rust spots,stains and other factors,a method of steel scribed line extraction based on Gabor transform is proposed.Through Gabor transformation,the line structures of characteristic direction and specific frequency are filtered and presented.Based on these line structures,the position of the line to be tested in the image is calculated,and then the double edges of a line are fitted by RANSAC algorithm,so as to accurately obtain the center position of the line.Experiments show that the algorithm can recognize the damaged and rusty scribed lines robustly when the quality of scribed lines is poor,and improve the detection accuracy and the robustness of the whole measurement system effectively,which has technical reference value.

关 键 词：图像处理 GABOR滤波 RANSAC 鲁棒性 

分 类 号：TB9[一般工业技术—计量学]