基于机器视觉的铆接孔几何参数测量  

Measurement of Rivet Hole Geometric Parameters Based on Machine Vision

作　　者：郝博 徐新岩 闫俊伟 Hao Bo;Xu Xinyan;Yan Junwei(Key Laboratory of Vibration and Control of Aero-Propulsion System Ministry of Education,Northeastern University,Shenyang 110819,China;不详)

机构地区：[1]东北大学航空动力装备振动及控制教育部重点实验室 [2]东北大学秦皇岛分校控制工程学院

出　　处：《工具技术》2024年第3期131-137,共7页Tool Engineering

基　　金：装备预先研究领域基金(61409230125)。

摘　　要：为解决传统方法测量铆接孔几何参数效率低、准确性差等问题,提出基于机器视觉的铆接孔几何参数测量方法。该方法使用CCD相机采集孔的特征信息,通过灰度处理、双边滤波及直方图均衡化,降低颜色、噪声对图像的影响,使用粒子群算法优化Otsu双阈值分割提取感兴趣区域。使用Zernike矩亚像素边缘检测代替传统边缘检测算法,提高边缘检测精度,再通过形态学处理弥补像素损失。采用改进随机Hough变换(Improved Randomized Hough Transform,IRHT)提取特征,实现孔的中心坐标和半径测量,利用像素当量标定,将像素测量值转化为物理尺寸。经实验验证,该方法测量两孔间距误差小于2%,测量半径为2mm的铆接孔误差小于4%,优于质心法、圆拟合等传统测量方法。In order to solve the problems of low efficiency and poor accuracy of the traditional rivet hole geometric parameter measurement method,a method based on machine vision is proposed.The hole characteristic information is captured by CCD industrial camera.Gray processing,bilateral filter and histogram equalization are used to reduce the influence of color and noise on the image.Particle swarm optimization Otsu dual-threshold segmentation is used to extract the region of interest.Zernike moment subpixel edge detection is used to improve the edge detection accuracy instead of the traditional edge detection algorithm.The previous pixel loss is compensated by morphological processing.Improved Randomized Hough Transform(IRHT)is used to measure the center coordinate and radius of the hole by extracting the features of the hole.Finally,pixel equivalent calibration is used to convert the measured pixel values into physical dimensions.The experimental results show that the error of hole spacing measured by this method is less than 2%,and the error of the rivet hole with 2mm radius is less than 4%,which is better than the traditional detection methods such as centroid algorithm and circle fitting.

关 键 词：机器视觉 图像处理 亚像素边缘检测 改进随机Hough变换 

分 类 号：TG806[金属学及工艺—公差测量技术] TH131.1[机械工程—机械制造及自动化]