基于机器视觉的锚具夹片牙面缺陷检测方法  

A detection method for tooth surface defects of anchorage clip based on machine vision

作　　者：唐滔[1] 王健[1] 曾庆宁[1] TANG Tao;WANG Jian;ZENG Qingning(School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学信息与通信学院

出　　处：《桂林电子科技大学学报》2019年第2期124-129,共6页Journal of Guilin University of Electronic Technology

基　　金：国家自然科学基金(61461011);认知无线电与信息处理省部共建教育部重点实验室主任基金(CRKL160107);桂林电子科技大学研究生教育创新计划(2017YJCX16,2017YJCX20)

摘　　要：针对锚具夹片牙面缺陷人工检测不稳定、效率低、成本高的缺点,提出一种基于机器视觉的检测方法。根据生产实际搭建锚具夹片牙面缺陷检测系统,通过工业相机采集锚具夹片工件图像,再进行图像预处理消除图像噪声;对锚具夹片牙面进行定位并提取ROI图像,对ROI图像进行边缘检测,提取锚具夹片牙面的边缘;经中值滤波和形态学处理,提取烂牙、平牙、重牙和光板缺陷特征,根据缺陷特征进行缺陷识别和分类。实验结果表明,每片锚具夹片平均检测时间为0.2s,缺陷分类准确率可达94%,可实现锚具夹片牙面缺陷的高速、高精度检测,满足企业生产的自动化检测要求。Aiming at the instability, low efficiency and high cost of the artificial detection for tooth surface defects of anchorage clip, this paper proposes a detection method based on machine vision. According to the actual production, a defect detection system is designed and set up. The image of the anchorage work-pieces is acquired with an industrial camera, and the image noise is eliminated through the image prepossessing. Then the anchorage dental surface is positioned to extract ROI image on which edge detection is performed to abstract the edge of tooth surface of anchorage clip. After median filtering and morphological processing, defect features of rotten tooth, flat tooth, heavy tooth and light plate are extracted. The defects are identified and classified in the light of the features. The experimental results show that the average detection time of each piece of anchorage clip is 0.2 s, and the accuracy rate of defect classification can reach 94%. That can realize the real-time and high-precision detection of the tooth surface defects of anchorage clips, and meet the automatic detection requirements of production companies.

关 键 词：锚具夹片 缺陷检测 机器视觉 ROI 边缘检测 形态学处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]