材料科学
光学
方位角
目视检查
计算机科学
生产线
直线(几何图形)
椭圆偏振法
对比度(视觉)
人工智能
计算机视觉
物理
薄膜
机械工程
纳米技术
工程类
数学
几何学
作者
Akira Kazama,Takahiko Oshige
摘要
An in-line defect inspection technique using polarized images for steel strip surface is developed. In inspection for low contrast defects, excessive-detection will be caused by harmless patterns such as slight oil patterns, chemical liquid patterns, and other patterns. We have adopted quasi-ellipsometric method using polarized images of the target samples to obtain their ellipsometric parameters, and found that the ellipsometric characteristics of the defects and the harmless patterns differ from each other. Based on this finding, we have developed an inspection system utilizing three polarized images with different azimuth angles to discriminate defects from harmless patterns at a high- speed production line.
科研通智能强力驱动
Strongly Powered by AbleSci AI