摘要
针对传统疵病检测技术因无法量化疵病等级而产生的精度限制,提出基于红外光电传感器的工业设备元件表面疵病检测技术。采用帧差法提取表面疵病图像特征,根据阈值设定获得经过处理的二值图像,确定特征属性;基于子图像极坐标空间表示法、齐次坐标以及变换矩阵方程,使待拼接图像的坐标完成转换,选取一帧属性图像作为根节点并查找与其具有相关性的图像,进而完成下一层级的相关性图像查找,待图像拼接完全,通过计算连通区域的长宽比,得到疵病参数指标数据,根据类别的初步判定对疵病等级进行划分,完成元件表面疵病检测。仿真实验表明,所提技术可以精准检测出元件表面疵病,具有优秀的检测性能与理想的适用性。
Aiming at the accuracy limitation caused by traditional defects detection technology due to the inability to quantify the defects level,an infrared photo sensor based industrial device component surface defects detection technology is proposed.The frame difference method is used to extract the surface defects image features,and the processed binary images are obtained according to the threshold setting to determine feature attributes.Based on the subimage polar coordinate space representation,homogeneous coordinates and transformation matrix equations,the coordinates of the images to be stitched are made to complete the conversion.Selecting a frame of attribute image as the root node and finding the image with correlation,and then,the next-level correlation image search is completed.After the images are completely stitched,calculate the aspect ratio of the connected area to obtain the defects parameter index data.Classifying the defects level according to the preliminary judgment of category to complete the component surface defects detection.Simulation experiments show that the proposed technology can accurately detect component surface defects,and has excellent detection performance and ideal applicability.
作者
王莉
史艳霞
范其明
WANG Li;SHI Yanxia;FAN Qiming(Intelligent Manufacturing College,Tianjin Sino-German University of Applied Sciences,Tianjin 300350,China)
出处
《激光杂志》
北大核心
2020年第8期30-34,共5页
Laser Journal
基金
天津市教委科研计划基金项目(No.2017KJ040)。
关键词
红外光电传感器
工业设备元件
表面疵病
检测技术
infrared photoelectric sensor
industrial equipment component
surface defects
detection technology
