中文摘要：

介绍了数学形态学在图像边缘检测领域的应用,针对带钢表面缺陷种类多、边缘方向多且含有噪声的特点,通过分析、试验,确定采用修正的抗噪膨胀腐蚀型算法,构造多种结构元素,对带钢图像进行边缘检测,提取出了很好的边缘检测效果。与传统边缘检测算子在满足速度要求的情况下相比,此种多结构元素数学形态学方法,不仅能有效地检测出带钢表面缺陷的各个方向上的边缘,而且具有很强的抗噪性能,可以很好地检测出在强噪声下微小的缺陷边缘。检测结果明显优于传统边缘检测算子。

英文摘要：

It introduces mathematical morphology in image edge detection’s application,aiming at numerous kinds surface defect、many edge directions and many noise of strip steel,through analyses、experiments,determine to use correctional noise immunity dilation and erotion operator,and structure multi-structural elements,applied to edge detection of strip steel,extract well result of edge detection.Then contrast and analysis the result of traditional edge detection operators,the result shows that,this method of multi-structural elements mathematical morphology,not only detect all directions surface defect of strip steel effectively,but also own strong noise-immunity,could detect small defect edge under strong noise.The result of detect is obvious well than traditional operators.