磁柱表面缺陷边缘检测算法优化研究

Research on the Optimization of Edge Detection Algoritm for Magnetic Column Surface Defects

磁柱应用十分广泛,且表面缺陷对其性能影响极大。然而,国内外对磁柱表面结疤、黑片缺陷的边缘检测算法研究很少。该文根据磁柱表面结疤、黑片缺陷的特征,基于优化的Canny提出一种适用于磁柱表面结疤、黑片缺陷的检测算法。在非极大值抑制过程中,将4个标准梯度方向优化为4个方向区域,图像中任意像素点梯度方向不再受限于4个标准方向;在双阈值处理过程中,将图像灰度级为t的领域灰度设定为[-m,m],利用优化后的均值p(t)计算出相应的类间方差σ^(*2)(t)、最佳阈值t*,实现通过高低阈值自适应选取来保留磁柱表面缺陷图像真正的弱边缘。研究结果表明,与Sobel、Canny算法检测结果相比,优化的Canny算法检测结果的缺陷边缘清晰度更高、连续性更好,且能完整检测出缺陷边缘,检测效果更佳。

Magnetic column is widely used,and it's surface defects have a great impact on the performance.However,there is little research on the edge detection algorithm of magnetic column surface scar and black chip defects at home and abroad.Based on the characteristics of the magnetic column surface scab and black chip defects and the optimized Canny Algorithm,a detection algorithm for the magnetic column surface scab and black chip defects was proposed.In the process of non maximum suppression,the four standard gradient directions were optimized to four direction regions,so the gradient direction of any pixel in the image is no longer limited by the four standard directions.In the process of double thresholding,the gray level of the image was set to the field gray level,and the optimized Mean Value p(t)was used to calculate the corresponding Interclass Variance σ^(*2)(t)and Optimal Threshold,so the real weak edge of magnetic column surface defect image is preserved by adaptive selection of high threshold and low threshold.The study results show that,compared with the detection results of Sobel Algorithm and Canny Algorithm,the detection results of the optimized Canny Algorithm has higher definition and better continuity of the defect edge,and can detect the defect edge completely,and the detection effect is better.