光学技术, 2018, 44 (1): 41, 网络出版: 2018-02-01
基于多尺度数学形态学的连铸坯缺陷边缘检测方法
Research on surface defects of continuous casting billets based on mathematical morphology
数学形态学 连铸坯缺陷 多尺度 边缘检测 mathematical morphology surface defects of continuous casting slabs multi-scale edge detection
摘要
连铸坯的缺陷会对轧制工艺产生影响, 通过对连铸坯表面图像特征进行分析, 提出了一种基于多尺度数学形态学的连铸坯表面缺陷检测方法。该缺陷检测方法能够有效去除噪声, 并对水痕、氧化铁皮、渣痕等伪特征具有较好的抑制作用。实验结果表明, 与Sobel方法、Prewitt方法、Canny方法、Laplacian方法等边缘检测方法相比较, 该方法对缺陷边缘提取完整、平滑、定位准确, 识别准确率达96.37%。
Abstract
By analyzing the surface defects characteristics of continuous casting billets, a mathematical morphology detection method based on multi-scale element is proposed. Through comparing with the traditional edge detection methods, the mathematical morphology based on multi-scale element not only can filter the noise effectively, but also can delete the false characteristic of the cracks, scales and slag. The method is detected with the samples of industrial line. The result shows that the method can detect edge completely and location correctively, which is better than the method of Sobel, Prewitt, Canny and Laplacian. The recognition rate is 96.37%。
闫俊红, 李忠虎, 王志春. 基于多尺度数学形态学的连铸坯缺陷边缘检测方法[J]. 光学技术, 2018, 44(1): 41. YAN Junhong, LI Zhonghu, WANG Zhichun. Research on surface defects of continuous casting billets based on mathematical morphology[J]. Optical Technique, 2018, 44(1): 41.