【作者】 王磊； 徐科；

【Author】 Wang Lei;Xu Ke;National Engineering Research Center for Advanced Rolling Technology,University of Science and Technology Beijing;

【机构】 北京科技大学高效轧制国家工程研究中心；

【摘要】 现有表面缺陷在线检测系统一般采用两维灰度图像检测方式,在氧化铁皮等伪缺陷的干扰下,缺陷的误报严重。本文提出一种基于光度立体视觉的钢板表面三维缺陷在线检测方案,通过绿色、蓝色两台激光线光源从不同角度对带钢表面照明,用一台彩色线阵3CCD摄像机高速扫描带钢表面,从采集到的彩色图像中分离出绿光和蓝光通道,根据三维重建算法得到带钢表面的三维信息,根据表面三维信息检测缺陷,从而降低由于伪缺陷造成的误报率。通过对热轧带钢表面辊印和冷轧带钢表面压痕缺陷进行表面三维检测试验,结果表明该方法可以准确检测到带钢表面三维缺陷。更多还原

【Abstract】 2D detection with gray level images ispopularly employed in current surface inspection systems of steel strips, andleads to high false detection of defects because of interference of pseudo-defects, such as scales. A3 Ddefect detection technique of steel plates based on photometric stereo is developed. Steel strips are illuminated with green and blue linear lasers from different directions.One color line-scan camera with 3CCD isemployed to scan themoving surface of steels. Color imagescaptured by the CCD camera are separated into green channel and blue channel, which are used for 3D reconstruction of steel strips. Defects are detected according to 3D information of steel surfaces,and false detection rate because of pseudo-defects is reduced. The technique wasexamined withroll imprints of hot-rolled steels and indentations of cold-rolled steels, andthe results showed that 3D defects of steel stripswere detected accurately.更多还原