一种在线铝带材表面缺陷提取方法  

A Surface Defect Detection Method for Online Aluminum Rolling Strip

作　　者：罗新斌[1,2] 黄秀琴[2] 贾建昇 邢青青[2] 

机构地区：[1]上海交通大学航空航天学院,上海200240 [2]苏州有色金属研究院有限公司,江苏苏州215026

出　　处：《有色金属加工》2013年第5期44-46,65,共4页Nonferrous Metals Processing

摘　　要：本文提出一种对铝带材表面缺陷在线自适应提取方法:首先对由线阵传感器采集来的材料表面图像数据进行高斯平滑和针对目标的空域增强;然后采用基于直方图的一维自适应熵的方法对预处理后的图像进行缺陷目标分割;再用形态学中的闭运算去除目标内部的孤立点和噪声,改善区域的联通性,得到最终的缺陷图像。实验结果表明该方法是一种精度高、耗时少、有效的在线缺陷定位方法。Based on the integration of entropy theory and mathematics morphology, an adaptive surface defect detection method on line is presented for aluminum rolling strip. The collected image data of the material surface through the linear array of sensors is smoothed by the Gaussian Filters, and the object to be segmented is highlighted by spatial filter. Then an automatic threshold value is obtained by applying information theory to the histogram data for defect object segmentation and the final defect image is obtained by the morphological closing function, by which the isolated points and the noise in the object are removed and the region＇s connectivity is improved. The experimental results show that it is an accurate, less time - consuming, and efficient algorithm for online positioning defects.

关 键 词：缺陷分割 自适应阈值分割 表面质量检测 信息熵 形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]