轨道扣件的图像采集与振动补偿方法

Image Acquisition and Vibration Compensation Method of Track Fasteners

摘要:

基于车载图像采集系统进行轨道扣件检测是保障列车运行安全的一项关键技术，受运行过程中低频机械振动和外界光照的影响，常规的图像采集检测方法所得图像精度较低。为了解决该问题，提出一种基于机器视觉光源模型的振动补偿方法。通过建立线阵相机、激光光源和扣件区域的照明模型消除外界光照的干扰，并采用频域分析法和多传感数据融合技术，设计补偿算法消除振动信息的干扰。最后，通过搭建采集平台进行实验验证，结果表明该方法可以有效地减弱外界光照对成像质量的影响，并对±5°以内的图像偏移实现了修正。与未采用机器视觉光源模型的方法相比，扣件区域对比度提高10倍以上。

Abstract:

Track fastener detection based on vehicle image acquisition system is a key technology to ensure train operation safety. However, due to the effect of low-frequency mechanical vibration and external light, routine acquisition and detection method usually gets poor-quality image. To solve this problem, a vibration compensation method based on machine vision system model was proposed. A lighting model composed of linear array camera, laser light source and fastener area were first set up to eliminate interference from external light. Combined with frequency domain analysis and multi-sensor data fusion technology, the compensation algorithm was designed to eliminate the interference of vibration information. In the end, an experimental platform was set up for verification. Results show that the above method can not only reduce external light effect, but also correct image offset that is within ±5°. What’s more, the contrast of the fastener area is more than 10 times higher than before.