不依赖于边缘的TFT-LCD面板缺陷自动检测方法  

Automatic detection of TFT-LCD panel defects without edge-dependence

作　　者：严成宸 闫真真[1] 

机构地区：[1]合肥工业大学仪器科学与光电工程学院,安徽合肥230009

出　　处：《河南科技学院学报（自然科学版）》2017年第3期65-71,共7页Journal of Henan Institute of Science and Technology(Natural Science Edition)

基　　金：科技部重大仪器专项资助项目(2013YQ220749)

摘　　要：为了检测TFT-LCD面板上可能出现的点缺陷和线缺陷,同时为应对其粗糙面及透明、半透明材质在光线的散射和多重反射下缺陷成像边缘较模糊的问题,提出了一种不依赖于边缘的TFT-LCD缺陷机器视觉自动检测方法.由工业CCD相机采集到的图像,采用Wiener滤波实现图像去噪;构造实值Gabor小波滤波器实现纹理背景抑制;采用基于最大熵的阈值分割方法分割缺陷;最后提取缺陷参数.通过实验实现了对样本图像中缺陷的判断、定位、面积测算和形状确定,与已有机器视觉检测方法相比,算法能够显著提高检测效率,适合缺陷在线检测.In order to detect point and line defects which may occur on TFT-LCD panels,at the same time,under the light scattering and multiple reflections,fuzzy edge of defects was imaged by the rough surface and transparent or translucent material.To deal with them,an independent-of-edge machine vision method of automatic detection of TFT-LCD defects was proposed.The images obtained by industrial CCD camera were denoised by Wiener filtering;the background texture was suppressed by constructing real value Gabor wavelet filter;the threshold segmentation method based on maximum entropy was employed to segment defects.Finally,the defect parameters were extracted.The judgment,positioning,area measuring and shape confirmation of defects in sample images were achieved through the experiment.Compared with the existing machine vision detection method,this algorithm can significantly improve the efficiency of detection,and is suitable for online detecting defects.

关 键 词：缺陷检测 WIENER滤波 GABOR滤波 最大熵分割 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]