Author Affiliations
Abstract
1 Anhui Province Key Laboratory of Measuring Theory and Precision Instrument, School of Instrument Science and Opto-electronic Engineering, Hefei University of Technology, Hefei 230009, China
2 Key Laboratory of High Power Laser and Physics, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, China
Fabrication of high-quality optics puts a strong demand on high-throughput detection of macroscopic bulk defects in optical components. A dark-field line confocal imaging method is proposed with two distinct advantages: (i) a point-to-line confocal scheme formed by a columnar elliptical mirror and an optical fiber bundle breaks through the constraint on light collection angle and field of view in the traditional line confocal microscopy using an objective, allowing for an extended confocal line field of more than 100 mm while maintaining a light collection angle of 27°; (ii) the bulk defects are independently illuminated as a function of time to eliminate the cross talk in the direction of the confocal slit, thus preserving point confocality and showing the optical section thicknesses to be 162 µm in the axial direction, and 19 and 22 µm in the orthogonal transverse directions. The experimental results verify that the method has a minimum detectable bulk defect of less than 5 µm and an imaging efficiency of . The method shows great potential in high-throughput and high-sensitivity bulk defects detection.
line confocal imaging dark-field imaging bulk defects detection Chinese Optics Letters
2023, 21(4): 041203
1 合肥工业大学仪器科学与光电工程学院, 安徽 合肥 230009
2 河南农业大学机电工程学院, 河南 郑州 450002
以智能制造业表面缺陷在线自动检测为应用背景,系统地综述了自动光学(视觉)检测(以下统称自动光学检测,AOI)技术。内容涉及AOI技术的基本原理、光学成像方法、系统集成关键技术、图像处理与缺陷分类方法等。对AOI系统集成中的关键技术,如视觉照明技术、大视场高速成像技术、分布式高速图像处理技术、精密传输和定位技术和网络化控制技术等进行了概述;对表面缺陷AOI主要光学成像方法的基本光学原理、功能和应用场合进行了总结;对表面缺陷检测中的图像处理、缺陷几何特征定义、特征识别与分类算法进行了系统阐述,重点介绍了周期纹理表面缺陷图像中的纹理背景去除方法,复杂和随机纹理表面缺陷的深度学习检测、识别与分类方法。
机器视觉 表面缺陷 自动光学检测 视觉检测 图像处理 分类
合肥工业大学仪器科学与光电工程学院，合肥 230009
为了检测 TFT-LCD平板表面缺陷，搭建了自动光学检测 (AOI)实验系统，提出了一种基于二维 DFT的缺陷检测算法。通过霍夫变换 (HT)检测到代表线状纹理的高能量频域直线，将位于直线邻域内的频率成分置 0，再经过二维 IDFT进行空间域图像重构，来移除方向性线状纹理背景，最后经过简单的阈值算法将缺陷从背景中提取出来。在对高能量频域直线的斜角计算时，利用数学统计解决了如何自动选取频域中高能量阈值以及如何从斜角直方图中自动提取各个峰值点这两个关键问题。在重构图像时，在以傅里叶频谱中心为圆心的圆环内，设置圆环内的频率成分不变，很好地保护了方向接近方向纹理的缺陷不被移除。实验结果表明，对包含纤维、污点和划痕的 TFT-LCD平板表面缺陷，检测结果完全正确，证明了该检测算法具有良好的实用性与鲁棒性。
缺陷检测 二维 DFT 线状纹理 霍夫变换 TFT-LCD TFT-LCD defect detection two-dimensional DFT linear texture Hough transform
