基于背景建模的钢轨表面缺陷像素级检测方法

钢轨表面缺陷具有独特性和稀疏性,利用机器视觉技术自动地检测钢轨表面缺陷仍存在很大挑战。提出一种基于背景建模的钢轨表面缺陷像素级检测方法,利用钢轨图像固有特性构建图像背景分布模型,找到背景分布簇中心,以定位到可疑像素点;提出一种钢轨表面缺陷像素级识别方法,根据可疑像素点的上下文特征和空间位置先验概率识别该像素点是否属于真实缺陷,并在钢轨缺陷数据集和实际线路上进行试验验证。研究结果表明:该方法在重载铁路和客运铁路2种钢轨缺陷数据集上均取得良好的识别性能,并在实际线路上达到100%的检测率。

Pixel-level detection method of rail surface defects based on background modeling

Rail surface defects are unique and sparse.It is still a great challenge to detect rail surface defects automatically by machine vision technology.A pixel-level detection method of rail surface defects based on background modeling was proposed,which uses the inherent characteristics of rail surface image to construct the image background distribution model,and finds the center points of background distribution cluster to locate abnormal pixel points.A pixel level recognition method of rail surface defects was proposed,which can identify whether the abnormal pixel point is a real defect region according to its context feature and spatial position prior probability.The rail defect datasets and the actual railway line are used to test the detection performance of the proposed method.The experimental results show that the proposed method has achieved good detection performance on two kinds of rail defect datasets and 100%detection rate on the actual railway line.