摘要
为了实现对触点零件的外观形貌质量检测,研究并提出了一种适用于机器视觉检测技术的触点类零件在线检测算法。针对触点零件几何及纹理缺陷的检测要求,设计了独特的算法流程。根据触点类零件的几何缺陷对零件图像提取了5种几何特征,针对零件图像的纹理缺陷,采用灰度共生矩阵的方法提取4种零件纹理特征,检测算法分步、有针对性的对触点进行几何及纹理缺陷的检测。文中对特征提取算法的有效性与常规的特征提取算法进行了对比试验,试验表明提出的算法流程较为简单实用,并能准确高效地识别出目前触点零件中存在的主要质量缺陷。系统基本满足触点零件在线检测的准确性及实时性要求。
In order to realize the quality detection for contact component＇s appearance, an online detecting system based on machine vision inspection technology is established. According to the detecting requirement of geometry and texture defect in contact components, a special software algorithm for this system are presented. In order to detect geometry and texture defect accurately, five kinds of geometry features and four kinds of texture features which are based on Gra^Level Cc^Occurrence Matrices method are extracted in accordance with defect types of contact components. The comparative experiment Which carries out to testify the effectiveness of feature extracting algorithms, indicates that these methods are simple and practical, resulting in correct and efficient identifications to the main quality defect of present contact components. It can satisfy the online detection system＇s requirements of real time and accuracy.
出处
《电子测量技术》
2010年第2期127-131,共5页
Electronic Measurement Technology
关键词
在线检测
触点零件
几何特征
纹理特征
On line detection
Contact component
Geometry defect
Texture defect
