基于大尺度三维几何视觉的货运列车超限检测方法研究

货运列车的超限是威胁铁路安全的重要因素之一。但是一直以来,我国货运列车超限检测都是采用落后的人工方式进行测量。提出了一种基于大尺度三维几何视觉的货运列车超限自动检测方法。该方法根据三维几何视觉确定被测物在世界坐标系下的坐标,并利用铁轨靶标得到铁轨坐标系和世界坐标系之间的转换关系,将被测物的坐标转换到铁轨坐标系中对货运列车是否超限做出判断。其中,外极线约束和结构光光条约束用来防止特征点的匹配错误,平面靶标用来计算相机的参数。外场实验结果表明,该方法可以快速、准确地对货运列车超限进行判断,具有很高的应用价值。

Freight train gauge-exceeding detection based on large scale 3D geometric vision measurement

Freight train gauge-exceeding is the main threat to railway safety.However,most railway stations in China are still using dated manual detection method nowadays.This paper describes a kind of freight train gauge-exceeding auto detection method based on large scale 3D geometric vision measurement.We reconstruct the 3D space model of the train under world coordinate system through stereo images,and judge if the train is gauge-exceeding or not with coordinate transition,which uses the relationship between rail coordinate system and world coordinate system using a rail target.Epipolar and structure laser stripe constraints are introduced to prevent the match error and plane targets are introduced to calculate the camera parameters.A prototype has been developed and experimental results show that the proposed method can detect freight train gauge-exceeding with high accuracy and speed.