小模数蜗杆外观缺陷的识别与分类

2018-09-21

作者：辛露1，2，李睿1，2 单位：1北京市精密测控技术与仪器工程技术研究中心；2北京工业大学

摘要：小模数蜗杆表面缺陷种类多，缺陷形状及尺寸大小差异较大，目前大多采用人工检测，效率较低。本文研发了一套基于机器视觉的检测系统，采用线阵相机扫描蜗杆圆周获得其表面图像，通过图像分割和形态学处理得到蜗杆缺陷形态。根据缺陷特点选取9种不同特征参数，使用高斯核函数建立支持向量机分类模型，实现了蜗杆缺陷的自动化检测，同时对其缺陷进行分类。试验结果表明，该方法检测准确率高，对工业生产中蜗杆表面质量评价具有实用价值。

关键词：视觉检测；形态学方法；特征提取；SVM

中图分类号：TG84；TH162文献标志码：A

Identification and Classification of Appearance Defect of Small Module Worm

Xin Lu，Li Rui

Abstract:There are many kinds of surface defects of small module worm，shape and size varied in different worms.Most of them are manual detection，the efficiency is low.To solve these problems，a machine vision based detection system is developed.To obtain the surface image using linear array camera scanning worm surface，through image segmentation and morphological processing worm defect morphology.According to the defect characteristics，selected 9 kinds of different characteristic parameters，using the Gauss kernel function to establish the classification model of support vector machine，achieve automatic detection of worm defects，and classify the defects.The experimental results show that this method has high detection accuracy and practical value of the worm surface quality in industrial production evaluation.

Keywords:visual inspection；morphology method；feature extraction；SVM