摘要:
针对带钢数字图像特点设计了一种表面缺陷检测算法, 能够基本解决带钢有无缺陷的初检工作。它从Hu不变矩的平移、旋转、比例不变性切入, 结合二阶矩的物理特性, 并考虑到无缺陷图像灰度过渡平缓, 置乱前后不变矩差较小; 而有缺陷图像的缺陷部分被破坏, 其置乱前后不变矩差相对较大, 据此设计并实现此算法。通过试验证实, 该算法能够快速判别出有缺陷的带钢图像, 使初检和缺陷类型识别能够实现并行处理, 对提高带钢缺陷检测系统的整体效率具有重要意义。
Abstract:
Considering characteristics of steel strips, an algorithm of surface defects initial inspecting for steel strip is designed to solve whether steel strip contains defects or not. It can meet requirements of automatic inspection. Using difference of physical characteristics of second moment invariant , the aim of recognition for the defects in steel strips is accomplished, it is proposed that changes of strip images containing defects are sharper than those containing no defect after image blocking and block scrambling, because moment invariants keep invariant after the image is transformed via translation, scaling, and rotation. The experimental results demonstrate that the algorithm can recognize steel strip images containing defects, making initial inspection and recognition parallel, and it is meaningful to improve efficiency of surface defect inspection system of steel strip.
下载:
