摘要
地铁车辆速度是列控系统的一项重要参数,传统的地铁车辆测速方法存在一定的缺陷。文章利用图像处理中的光流法研究地铁车辆测速新方法,提出了满足地铁车辆测速实时性和精度要求的快速光流计算方法。引入图像金字塔,通过由粗到精、逐层细化的步骤求解大位移运动光流,解决了LK光流法无法计算大位移运动的缺陷;提出了将GPU用于光流计算,实现图像特征点检测计算并行化,显著提升光流计算速度;设计了线路仿真软件进行测速仿真试验。结果表明,基于图像金字塔的光流改进算法提高了大位移情况下光流估计精度,基于GPU的光流并行计算有效提高了光流计算速度,满足地铁车辆测速系统的实时性要求。通过搭建支撑平台,并进行实际线路试验,进一步验证了光流测速方法的可行性。
Subway speed is an important parameter of train control system.The traditional subway speed measurement methods have some defects.In this paper,a new method of subway speed measurement by using the optical flow method in image processing was explored,and a fast optical flow calculation method to meet the requirements of real-time and accuracy of subway speed measurement was put forward.The image pyramid was introduced to solve the optical flow of large displacement motion through the method of coarse to fine and layer by layer refinement,which solved the defect that LK optical flow method could not calculate large displacement motion;GPU was used in optical flow calculation to realize the parallelization of image feature point detection and calculation,which significantly improved the speed of optical flow calculation;The line simulation software was designed for speed measurement simulation experiment.The results show that the improved optical flow algorithm based on image pyramid can improve the accuracy of optical flow estimation under large displacement,and the optical flow parallel calculation based on GPU can effectively improve the optical flow calculation speed and meet the real-time requirements of metro speed measurement system.The feasibility of the speed measurement method in this paper is further verified by building the machine support platform and carrying out the actual line experiment.
作者
吕红强
黄涛
聂兴家
陈柯宇
蔡正凯
单奇
LYU Hongqiang;HUANG Tao;NIE Xingjia;CHEN Keyu;CAI Zhengkai;SHAN Qi(CRRC Nanjing Puzhen Co.,Ltd.,Nanjing,Jiangsu 210031,China;School of Mechanical Engineering,Southwest Jiaotong University,Chengdu,Sichuan 610031,China)
出处
《机车电传动》
北大核心
2022年第2期129-134,共6页
Electric Drive for Locomotives
基金
国家自然科学基金项目(51605393)
国家重大技术装备攻关工程项目(系列化中国标准地铁列车研制及试验)。
关键词
地铁
测速
光流法
GPU加速
实时性
仿真
metro train
speed measurement
optical flow method
GPU acceleration
real time
simulation
