高速铁路轨道板图像智能比对技术

Intelligent Image Comparison Technology for High Speed Railway Track Slabs

提出一种结合深度学习与经典图像处理技术的高铁轨道板外观状态变化智能比对方法,以扣件位置为锚点对齐轨道板,通过网格化图像提取特征,分区域比对轨道板图像相似性。利用同一块轨道板的历史图像数据与新采集的数据自动进行比对,寻找外观差异,从而在无需预先定义故障类型的情况下筛出轨道板异物、设备形位变化及破损等故障,提高了检测效率和检测精度,进一步保障高铁安全运营。

An intelligent method to compare changes in appearance status of high speed railway track slabs by using the deep learning and classical image processing technology is proposed. Track slab is aligned with the anchor point set at the fastening position and characteristics are extracted through image grid to compare track slab images by areas and find out the similarity. The historical image data of a track slab is automatically compared with its newly collected data to find out the difference in the appearance, and thus identify foreign matters on the track slab, equipment geometrical changes, damage and other faults without defining the fault type in advance. In this way, the detection efficiency and accuracy are improved and safe operation of high speed railways is further ensured.