基于机器视觉的圆斑状钢轨擦伤检测算法

Detection Algorithm of Circular Spot Rail Scratch Based on Machine Vision

针对现有基于灰度阈值的钢轨擦伤检测算法受光照等外部环境的影响较大的问题,本文提出了一种基于机器视觉的圆斑状钢轨擦伤检测算法。首先通过分析采集图像在垂直方向的灰度均值曲线,提取出钢轨顶面区域;然后运用边缘检测的方法得到擦伤区域边缘的候选像素点;最后运用形态学处理删除不属于擦伤区域的虚假边缘,确定钢轨擦伤区域的位置。用测试数据集对本文算法进行检测性能评测,并与基于灰度阈值的算法进行对比。结果表明:本文算法对圆斑状钢轨擦伤样本的检测准确率为96.4%,而基于灰度阈值的算法的检测准确率为86.8%,本文算法的检测准确率大幅提升,能够对钢轨擦伤进行有效检测。

Aiming at the problem that the existing rail scratch detection algorithm based on gray threshold is greatly affected by the external environment such as light,this paper proposed a circular spot rail scratch detection algorithm based on machine vision.Firstly,the rail top area was extracted by analyzing the gray mean value curve of the collected image in the vertical direction.Then the candidate pixel dots of the edge in the scratch area were obtained by edge detection.Finally,the false edges that do not belong to the scratch area were deleted by morphological processing to determine the location of the rail scratch area.The test data set was used to evaluate the detection performance of the algorithm presented in this paper,and compared with the algorithm based on gray threshold.The results show that the detection accuracy of the algorithm in this paper is 96.4% for the circular spot rail scratch samples,while the detection accuracy of the algorithm based on the gray threshold is 86.8%.The detection accuracy of the algorithm in this paper is greatly improved,which can effectively detect the rail scratch.