基于2D的钢轨轮廓特征点提取方法研究

摘要

Rapid and accurate extraction of feature points of rail contour data is the prerequisite to ensure accurate matching of rail contour and detection of track geometric irregularity. This paper conducts a study on the extraction method of the rail profile measurement data features based on the two-dimensional laser displacement sensor(2D). By using smoothing filtering method based on mid-value error and continuous self-adaption adjusting value, the paper carries out the smoothing filtering of the measured profile data to allow the sectional profile data being sectionally smoothed. The concept of rail profile feature curve is put forward, a definition method of feature curve is given, and measured profile feature points are rapidly located based on feature points on the feature curve. Finally,GJ-2 track inspection car is employed to measure the actual track profile with the feature point extraction method proposed in this paper to extract the measured data feature points. The test results show that this method can locate feature points quickly and accurately.%钢轨轮廓数据特征点快速、准确提取是保证钢轨轮廓精确匹配、轨道几何不平顺精确检测的前提.对基于二维激光位移传感器(2D)的钢轨轮廓测量数据特征点提取方法进行研究,通过对采集的钢轨半断面轮廓数据采用基于中值误差与连续度自适应调整权值的平滑滤波方法对实测轮廓数据进行平滑处理,解决存在分段的轮廓数据达到分段平滑的效果.提出钢轨轮廓特征曲线的概念,并给出特征曲线的一种定义方式,利用特征曲线上的特征点去快速定位实测轮廓特征点.最后,采用GJ-2型轨道检测车进行试验,通过对实际轨道进行轮廓测量,采用本文所提出的特征点提取方法对实测轮廓数据进行特征点提取,试验证明,该方法能快速、准确地定位轮廓特征点.