小型平板玻璃表面缺陷检测系统  

Surface defect detection system for small-scale flat-panel glass

作　　者：李长有[1] 刘遵[1] 李帅涛[1] 

机构地区：[1]河南理工大学机械与动力工程学院,河南焦作454003

出　　处：《计算机应用》2017年第A02期167-170,182,共5页journal of Computer Applications

基　　金：河南省教育厅科学技术研究重点项目(14A460002)

摘　　要：搭建了一种低角度照明的小型平板玻璃缺陷检测机器视觉系统,针对玻璃缺陷图像存在的不均匀光照干扰,采用一种基于线扫描灰度波动变换的自适应阈值分割方法。首先,在水平和竖直两个方向上,分别通过"弓"字形线扫描把图像变成两条灰度波动曲线;然后,通过灰度波动变换将两条曲线还原成两幅图像;接着,将两幅图像重构为一幅新图像;最后,用Otsu法分割新图像。实验结果表明,该方法可以有效地减弱不均匀光照对玻璃缺陷检测系统的影响,提高玻璃缺陷的分割效果。A machine vision system with a low angle illumination for small flat glass defect detection was established, and an adaptive threshold segmentation algorithm based on line scan gray-scale fluctuation transformation was proposed for the nonuniform illumination existed in glass defect image. Firstly, In the horizontal and vertical directions, the image was turned into two gray-scale fluctuation curves through U-shaped reciprocating scan. Then, the two curves were transformed into two images by the gray-scale fluctuations transformation, afterwards the two images were reconstructed into a new image. Finally, Otsu method was used to segment the new image. The experimental results show that, the algorithm can effectively reduce the influence of uneven illumination on the glass defect detection system and improve segmentation effects of glass defects.

关 键 词：玻璃表面缺陷 图像分割 机器视觉 灰度波动 自动化检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]