近红外钢轨磨耗检测中的图像处理
Image Processing of Near-Infrared Rail Abrasion Detection System
-
摘要: 对近红外钢轨磨耗检测中实时图像处理方法进行了研究.在断面磨耗检测中,要获取较高精度的磨耗数值,起始点(或称坐标基准)的定位必须准确且误差小.考虑到标准钢轨轨形和实际钢轨磨耗情况,得出了起始点两种可行的定位方式,并就其中的一种用编程实现,仿真处理效果良好,所获取的两关键点的磨耗数值与实测值吻合较好.为了进行实采钢轨轨形与标准轨形的模板比较,讨论了钢轨断面图像的细化问题.这里并没有采用通常的细化算法,而是先利用Roberts算子进行图像的边缘检测,再抽取出光带中心线,大大减少了运算量,提高了运算速度,同时也满足了对轨形细化的要求.断面图像的起始点定位与轨形细化兼顾了处理精度和速度的要求,满足了本系统实时检测的要求.Abstract: The methodology of real time image processing of near-infrared rail abrasion detection system has been studied. In consideration of accurate abrasion value,the location of starting point is very important. For this, two feasible location methods have been offered. One of the two methods has been put into practice. And the results are in line with the actual abrasion values. In addition, the key technology of the extraction of reflection-streak center line has been discussed. For more,Roberts edge detection has been applied and the single-pixel center line is extracted. The method got a good effect and their processing speed is fast,which show that it can meet the demand for real time cross-section detection system.