基于机器视觉的玻璃缺陷检测

随着科学技术的进步,高端显示屏产品对平板玻璃的质量要求越来越高,玻璃的表面缺陷检测技术也因此备受关注。传统的人眼检测方法工作量大且准确率低,已经无法满足生产实际要求。研究了一种基于机器视觉的玻璃质量检测系统,采用先进的CCD成像技术和背光式照明获取图像,用MATLAB图像处理工具箱对采集到的图像进行灰度值化、滤波降噪和阈值分割处理,实现对缺陷区域的特征提取和识别。最后用BP神经网络对玻璃表面的三种缺陷进行分类,该神经网络识别的平均误差率为9.84%,表明此检测方法具有一定的应用价值。

Glass defects inspection based on machine vision

With the improvement of science and technology, the advanced display products have a higher requirement on the quality of flat glass. Therefore, the surface detection technology of the glass defects has attracted a great attention. The traditional inspection method for the surface defects is mostly conducted by naked-eye observing. This manual inspection method is time-consuming, with low accuracy, and unable to meet the production requirement. An inspection system for the glass based on machine vision was studied in this paper. A CCD camera and a backlit illumination were used to acquire the images of glass. With the help of the powerful image processing toolbox of MATLAB software, a sequence of operations was performed on the image to achieve the feature extraction and the defect recognition, such as grayscale transformation, denoising, and threshold segmentation. The BP neural network was built to classify three kinds of defects on the glass surface and the results show that the average error rate of this neural network''s recognition is 9.84%. This inspection method has a wide application.