基于图像处理的潘得路快速弹条扣件状态识别

State Recognition of Pandrol Fast Projectile Fastener Based on Image Processing

为研究针对潘得路快速弹条扣件状态识别的方法,通过一种基于外触发模式同步控制线阵相机的方法,以保证线阵相机能够在列车变速运行情况下采集的图像不产生纵向畸变;对采集到的图像进行预处理,利用十字交叉法实现扣件区域的定位并提取;采用基于快速归一化积相关算法的模板匹配定位扣件区域的扣件绝缘帽、固定螺栓并获取其中心坐标参数,在此基础上提出一种基于特征距离的扣件状态识别方法。通过搭建图像采集轨道车实验验证了该同步控制法能够保证线阵相机在列车变速运行时采集的图像不产生纵向畸变;进一步实验验证该快速归一化积相关算法相较于归一化积相关算法匹配速度提升了近60%,同时实验验证了该基于特征距离的扣件状态识别方法能够实现潘得路快速弹条扣件状态的识别。

In order to detect the state of the Pandrol fast projectile fastener,a method of synchronous control of linear camera based on external trigger mode is used to ensure that the fastener’s images be collecting undistorted when the train’s speed changes.Preprocessing the collected images,and using cross crossing method to locate and extract fastener area.A template matching method based on fast normalized cross correlation algorithm is used to locate the fastener insulation cap and fixed bolt position in fastener area and obtain its central coordinate parameters,then,a recognition method of fastener status based on feature distance is proposed.The experiment of image acquisition railroad car proves that the synchronization control method can ensure that the images collected by linear array camera do not produce longitudinal distortion when the train is running at variable speed;Further experiment shows that the matching speed of the fast normalized cross correlation algorithm is improved by nearly 60%compared with the normalized cross correlation algorithm;At the same time,the experiment proves that the method based on feature distance can recognize the state of the Pandrol fast projectile fastener.