基于形态学玻璃屏幕表面划痕检测方法研究  

Study on scratch inspection methods of glass screen surface based on morphology

作　　者：赵文宏[1] 彭超[1] 陈红星[1] 

机构地区：[1]浙江工业大学机械工程学院,浙江杭州310014

出　　处：《浙江工业大学学报》2016年第3期279-282,共4页Journal of Zhejiang University of Technology

基　　金：国家自然科学基金面上项目(51275476)

摘　　要：针对玻璃屏幕表面划痕检测的要求,进行了基于形态学玻璃屏幕表面质量划痕检测方法的研究.通过激光散射扫描获取图像,利用均值滤波平滑图像,采用局部动态阈值分割获取二值图像,然后计算所有相连接的区域剔除干扰部分,再利用膨胀、骨架提取等形态学处理方法提取划痕缺陷,最后实验测量了大量的划痕样本的特征值,并确定划痕判据.为了验证检测方案的可行性进行了实验,实验结果表明:划痕识别率高达96%,误判率3%,漏检率为1%.所提出的检测方法及判据能够准确地识别玻璃划痕缺陷,同时具有检测速度快的优点,满足了玻璃屏幕表面划痕的检测要求.This paper studied surface quality detection technology of glass screen based on morphology methods,according to the scratches detecting requirements of glass screen surface.Firstly,the requested images are acquired by laser scatter scanning,using filter smoothing image,to get the binary image by local dynamic threshold segmentation.Then we calculated the area of all the connected parts to remove the interference parts,after that morphological method was used to extract scratches defects,such as the expansion and skeleton extraction.Finally,the scratch criterion were acquired by calculating the eigenvalues of many scratch samples.To verify the feasibility of the detection scheme,this paper carried out experiments.The experimental results showed that the detection methods and criteria of the proposed can accurately identify glass scratch defects,which meets the requirements of the glass screen surface scratch detection.Scratch recognition rate reached 96% and 3% error rate,false negative rate was 1%.

关 键 词：划痕检测 激光散射 图像处理 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]