基于图像技术的无砟道床表观伤损检测系统研制

Manufacture of Apparent Damage Detection System for Ballastless Bed Based on Image Technology

我国高速铁路部分线路无砟道床裂缝、离缝、缺损等表观伤损主要依赖于人工持钢板尺、塞尺、测宽仪等传统工具进行测量,检测效率低,准确性差。针对这一问题,本文基于图像技术研制了无砟道床表观伤损检测系统,对无砟道床表观高清图像快速采集,利用图像处理、深度学习等技术构建图像识别算法,编制识别软件对采集图像进行智能识别。现场验证结果表明,该系统可大大提高检测效率,系统精度达到0.1 mm,识别准确率超过90%。

The apparent damage of ballastless bed of some high speed railways in China,such as crack,gap and defect,mainly depends on traditional tools such as steel ruler,feeler gauge and width gauge,which has low detection efficiency and poor accuracy.To solve this problem,the apparent damage detection system of ballastless bed based on image technology was developed in this paper,which can quickly collect the apparent high-definition image of ballastless bed,construct the image recognition algorithm by using image processing,deep learning and other technologies,and compile the recognition software to intelligently recognize the collected image.The field verification results show that the system can greatly improve the detection efficiency,the system accuracy reaches 0.1 mm,and the recognition accuracy exceeds 90%.