摘要
针对当前铁路钢轨检测算法在识别中准确性和鲁棒性不高的问题,提出采用Hu不变矩特征实现轨道线搜索,并以B样条曲线为拟合模型的钢轨自动检测方法。算法根据视频帧中钢轨的边缘特征,通过改进的霍夫变换识别并确定图像空间的轨道线消隐边界,完成近远景区的标定。针对近景区直轨,通过直线模型拟合;在远景区,采用可漂移检测窗通过比对Hu不变矩来提取轨道特征点,以最小二乘法实现B样条曲线模型拟合。并制定模型更新和切换原则自动跟踪轨迹线。实验结果表明:轨道线平均跟踪时间为0. 081 s,可以提高钢轨检测识别的精确性和鲁棒性,能够更好地解决曲线轨道的模型拟合问题。
To address the problem of low accuracy and robustness in the existing railway track detection algorithm in recognition,an automatic rail inspection algorithm based on Hu invariant moments was proposed,which can be used to search the track line and which used B spline curve as the fitting model.Firstly,based on the edge feature of the rail in video frames,the algorithm was used to recognize the position of the track by the improved Hough transform,to determine the hidden boundary of the track line and to complete the calibration of the close and far range areas.For straight track in close range area,the linear model was used to calibrate the front straight rail.In the far range area,feature points were extracted by searching the Hu invariant moment feature based on the drift window search matching algorithm.Then,the least square method was used to fit B spline curve model.Finally,based on the model switching and the window searching strategy,the track line was tracked automatically.The experimental results show that the average time of tracking of the track line is 0.081 s,and the algorithm can improve the accuracy and robustness of rail detection and better solve the problem of the model fitting of curved track.
作者
董昱
郭碧
DONG Yu;GUO Bi(School of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;Key Laboratory of Opto-technology and Intelligent Control of the Ministry of Education,Lanzhou Jiaotong University,Lanzhou 730070,China)
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2018年第10期64-70,共7页
Journal of the China Railway Society
基金
国家自然科学基金(61763023)
关键词
列车前方环境理解
铁轨检测
Hu不变矩特征
可切换曲线模型
train front environmental understanding
rail detection
Hu moment invariant features
switching curve model
