基于机器视觉的铁路轨道监测方法

A railroad track checking strategy in light of machine vision

随着各地城际铁路、市域铁路、轨道交通、公路、管网的建设持续推进,施工建设大量出现在既有铁路附近,对既有铁路、高速铁路安全运营产生巨大风险。传统解决方案通过人工监测或半自动化监测,方法效率低、监测难点多、管理系统化程度低,不能实时反馈轨道变形数据,影响施工决策。针对上述不足,本文基于机器视觉相机成像模型与三维信息恢复理论方法,采用亚像素插值算法,联合拟合计算出不同期次钢轨轨面中心点位置,同时获得轨道中心点成果精度评估。通过轨道施工现场的试验和结果分析表明,本文算法可有效提取轨道中心点信息,为铁路运营安全监测提供准确的轨道中心点变化数据。

With the consistent advancement of the development of intercity rail lines,metropolitan rail routes,rail travel,streets and line organizations,an enormous number of development projects are close to the current railroads,which could present incredible dangers to the protected activity of the current rail lines and fast rail routes.Through manual checking or self-loader observing,the conventional arrangement has low productivity,many checking troubles and low level of systematization of the board,and it can not criticism the track disfigurement information continuously and influence the development independent direction.Taking into account the above inadequacies,in light of the machine vision camera imaging model and the hypothetical technique for three-layered data recuperation,in this paper,the sub-pixel introduction calculation was utilized to together fit and compute the place of the middle mark of the rail surface in various periods,and get the exactness assessment of the consequence of the middle mark of the rail simultaneously.The test and result examination on the track building site showed that the calculation in this paper could successfully remove the track place data and give precise track community change information for railroad activity security observingn.