摘要
针对冷轧带钢表面缺陷在在线检测过程中,无法准确地检测到图像中缺陷的边缘问题,提出了一种基于小波变换模极大值的板带钢表面缺陷多尺度边缘检测算法.该方法较好地解决了边缘检测精度与抗噪性能之间的协调问题,实现了在多个尺度上板带钢表面缺陷的边缘提取.实验结果表明该方法对伪缺陷边缘的去除有很好的效果,同时能够较好地保留图像中缺陷的边缘细节信息,具有更好的边缘检测性能,为带钢表面缺陷在线检测系统中的后续处理,如图像自动分割、缺陷识别等奠定了基础.
The image edge of steel strip surface defects cannot be detected exactly on-line at present.It is a serious problem to be solved.A wavelet-based maximum modulus algorithm is therefore proposed for edge protection,which can coordinate well the edge detection precision and denoising effect to extract the surface defect edges on multi-scale.Experimental results showed that the new algorithm can exclude well the fault detect edge from detection with the particulars of defect edge kept up,thus providing better edge detectivity for the subsequent online processing in surface defect detection,such as image auto-segmentation,defect identification,etc.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第3期432-435,共4页
Journal of Northeastern University(Natural Science)
基金
国家自然科学基金资助项目(50574019)
国家高技术研究发展计划项目(2008AA04Z135)
关键词
边缘检测
图像处理
表面缺陷
多尺度分析
小波变换
edge detection
image processing
surface defect
multi-scale analysis
wavelet transform
