非接触红外光电传感器在强反射金属缺陷检测中的应用  

Application of Non-Contact Infrared Photoelectric Sensor in the Detection of Strong Reflection Metal defects

作　　者：何成 马思远 周武杰[1] HE Cheng;MA Siyuan;ZHOU Wujie(School of Information and Electronic Engineering,Zhejiang University of Science and Technology,Hangzhou Zhejiang 310023,China;College of Control Science and Engineering,Zhejiang University,Hangzhou Zhejiang 310027,China)

机构地区：[1]浙江科技学院信息与电子工程学院,浙江杭州310023 [2]浙江大学控制科学与工程学院,浙江杭州310027

出　　处：《传感技术学报》2023年第9期1497-1502,共6页Chinese Journal of Sensors and Actuators

基　　金：浙江省自然科学基金项目(LQ19F010001)。

摘　　要：针对光照时强反射金属表面大量的自由电子吸收所有频率的光,导致金属表面缺陷特征弱化,降低金属表面缺陷检测准确率的问题,提出非接触红外光电传感器在强反射金属缺陷检测中的应用研究。首先,利用红外光电传感器获取强反射下的金属表面红外图像,并将红外成像频率因子引入原有直方图中,形成金属表面加权直方图;然后,均衡化处理金属表面直方图中的红外图像像素,消除强反射带来的光学干扰;最后,利用灰度共生矩阵和卷积神经网络的方法,检测及分类红外图像进行金属表面特征提取,实现强反射下金属表面缺陷准确检测。实验结果表明,对于不同类别的金属表面缺陷,所提方法的准确率均高于96.4%,检测耗时均低于9.6 s,可以快速准确完成金属表面缺陷检测,具有普遍适应性。In view of the problem that a large number of free electrons on the strongly reflective metal surface absorb light of all frequencies,weakening the characteristics of metal surface defects and reducing the accuracy of metal surface defect detection,the application of non-contact infrared photoelectric sensor in the strongly reflective metal defect detection is proposed.Firstly,the infrared image of the metal surface under strong reflection is obtained by using the infrared photoelectric sensor,and the infrared imaging frequency factor is introduced into the original histogram to form the weighted histogram of the metal surface.Then,the infrared image pixels in the histogram of the metal surface are equalized to eliminate the optical interference caused by strong reflection.Finally,the gray level co-occurrence matrix and convolution neural network are used to detect and classify the infrared images for metal surface features,and the metal surface defects under strong reflection are accurately detected.The experimental results show that for different types of metal surface defects,the accuracy of the proposed method is higher than 96.4%,and the detection time is less than 9.6 s,which can quickly and accurately complete the metal surface defect detection,and has universal adaptability.

关 键 词：红外光电传感器 强反射 金属表面 红外成像频率因子 灰度共生矩阵 卷积神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]