Abstract: The image processing algorithms are introduced in the abrasion detection of locomotive pantograph slide, and a combined method is proposed to detect the actual thickness of pantograph slide. The edge detection for images is always susceptible to noise, uneven illumination and other factors. The Hough transform is employed for image horizontal correction. To reduce the influence of noise and uneven illumination, the fuzzy clustering algorithm is used to segment image, and the segmented image is edge-detected by Canny operator to extract the edge information of pantograph slide. Because the extracted edges are of discontinuity, an edge growing method combined with morphology is used to connect the discontinuous lines. The edge lines are calibrated to calculate the abrasion of pantograph slide.

Key words: image processing, pantograph slide, fuzzy clustering, edge detection, edge growing

摘要： 将图像处理算法引入机车受电弓滑板的磨耗检测中，提出一种检测受电弓滑板实际厚度的测量算法。由于直接对图像进行边缘检测易受噪声、光照等因素影响，通过Hough变换对图像进行水平校正，由模糊聚类算法进行图像分割以去除光照不均及噪声的干扰，对分割后图像采用Canny算子精确提取滑板的边缘信息。考虑提取边缘的不连续性，引入结合数学形态学的边缘生长方法实现不连续边缘的连接，通过图像标定计算出实际滑板磨耗厚度。

关键词: 图像处理, 受电弓滑板, 模糊聚类, 边缘检测, 边缘生长