基于机器视觉的齿轮齿形缺陷检测技术  

Gear Tooth Defect Detection Technology Based on Machine Vision

作　　者：任聪 余昌水 Ren Cong;Yu Changshui(Shaanxi Fast Gear Co.,Ltd.,Shaanxi Xi'an 710119)

机构地区：[1]陕西法士特齿轮有限责任公司

出　　处：《汽车实用技术》2019年第16期95-97,共3页Automobile Applied Technology

摘　　要：文章主要面向齿轮工件,基于机器视觉方法实现齿形轮廓自动检测。首先设计齿形缺陷专用检测装置,用于齿轮图像的采集;然后基于机器视觉方法,对齿轮图像进行预处理,包括灰度转换、中值滤波去噪以及二值化;并基于Sobel边缘检测实现齿形轮廓的提取,确定分度圆半径,基于最小二乘法拟合圆求圆心,根据分度圆与齿廓的交点测量齿距偏差。经验证,齿轮图像预处理以及Sobel边缘检测具有较好的通用性,基于机器视觉的齿形缺陷检测准确可靠,能够有效提升齿轮轮廓的自动检测效率。This paper mainly focuses on gear workpiece,and realizes tooth profile detection based on machine vision.Firstly,a special detection device for gear profile defect is designed to collect gear image.Then,based on machine vision,gear image is preprocessed,including gray level conversion,median filter denoising and binarization.Then,tooth profile is extracted based on Sobel edge detection,the radius of indexing circle is determined,the center of circle is obtained by least square fitting circle,and the center of circle is obtained according to indexing circle and tooth profile.Measure pitch deviation at intersection point.Experiments show that gear image preprocessing and Sobel edge detection have good versatility.The detection of tooth profile defects based on machine vision is accurate and reliable,and can effectively improve the efficiency of automatic detection of gear profile.

关 键 词：齿形检测 机器视觉 缺陷检测装置 SOBEL边缘检测 

分 类 号：TG86[金属学及工艺—公差测量技术]