基于机器视觉的树脂镜片表面疵病边缘检测  

Edge Detection of Defect for Resin Lenses Based on Machine Vision

作　　者：林磊 臧铁钢[1] LIN Lei;ZANG Tiegang(College of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)

机构地区：[1]南京航空航天大学机电学院,江苏南京210016

出　　处：《机械制造与自动化》2021年第2期230-232,240,共4页Machine Building & Automation

摘　　要：针对树脂镜片表面疵病自动化检测的需求,开展了基于机器视觉的树脂镜片表面疵病边缘检测研究。设计一种改进的快速Hough圆检测方法获取镜片图像的感兴趣区域,采用混合滤波消除脉冲噪声和高斯噪声,提出一种基于局部窗口阈值的改进Canny边缘检测算法,选取局部最佳分割阈值,对窗口内的疵病实现精确分割。实验结果表明,该边缘检测算法能够较好地检测疵病边缘细节。To meet the needs of automatic detection of surface defect for resin lenses,research on edge detection of defect for resin lenses based on machine vision was conducted.An improved fast Hough circle detection method to obtain the region of interest of the lens image was designed,hybrid filtering was used to eliminate impulse noise and Gaussian noise,and an improved Canny edge detection algorithm based on the local window threshold was proposed to select the local optimal segmentation threshold and get accurate segmentation of defect in the local window.The experiment results show that this edge detection algorithm can better detect edges of defect.

关 键 词：机器视觉 树脂镜片 疵病检测 混合滤波 边缘检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]