首页 | 本学科首页   官方微博 | 高级检索  
     

道路监控系统中非同步相机时间校正方法
引用本文:梁华刚, 程加乐, 孙小喃, 茹锋. 道路监控系统中非同步相机时间校正方法[J]. 交通运输工程学报, 2015, 15(4): 118-126. doi: 10.19818/j.cnki.1671-1637.2015.04.015
作者姓名:梁华刚  程加乐  孙小喃  茹锋
作者单位:长安大学 电子与控制工程学院, 陕西 西安 710064
基金项目:国家自然科学基金项目61203374 陕西省科学技术研究发展计划项目2014KW01-05
摘    要:
为了提高交通事件还原的准确性, 提出一种基于双目极几何关系的高精度交通现场三维重建时间校正方法。分析了多个非同步相机帧序列, 建立了非同步相机时间校正模型。为了校正多部非同步相机的时间误差, 采用极几何约束和点到极线距离最小约束准则, 进行对应帧时间同步匹配。由2部非同步相机组成拍摄系统, 进行对应帧时间匹配误差分析, 估算视频序列起始帧时间误差, 实现高精度动态场景复原。通过闪烁LED灯的移动模拟高速移动的车辆, 估计时间误差范围。为了进一步验证方法在实际道路环境中的有效性, 利用4部相机拍摄自由落体的皮球视频序列, 分别使用相似度判别的时间校正方法和本文方法进行对比试验。试验结果表明: 当相机的曝光时间为1/30s, 高速移动的发光源闪烁周期为6.59ms时, 视频序列的起始帧时间误差范围为0~6.59ms, 在这个误差范围内的对应帧数达到88.1%以上; 与相似度判别的时间校正方法相比, 本文方法效果较好; 与几种传统时间校正方法相比, 本文方法显著提高了道路监控系统中非同步相机对应帧时间误差匹配和三维重建的精确度。

关 键 词:智能交通系统   道路监控系统   机器视觉   非同步相机   时间校正   极几何   对应帧匹配
收稿时间:2015-02-13

Time correction method of asynchronous cameras in road monitoring system
LIANG Hua-gang, CHENG Jia-le, SUN Xiao-nan, RU Feng. Time correction method of asynchronous cameras in road monitoring system[J]. Journal of Traffic and Transportation Engineering, 2015, 15(4): 118-126. doi: 10.19818/j.cnki.1671-1637.2015.04.015
Authors:LIANG Hua-gang  CHENG Jia-le  SUN Xiao-nan  RU Feng
Affiliation:School of Electronics and Control Engineering, Chang'an University, Xi'an 710064, Shaanxi, China
Abstract:
In order to improve the reduction accuracy of traffic incident, a high accuracy time correction method of traffic scene 3Dreconstruction was proposed based on binocular epipolar geometry relationship.The frame sequences of multiple asynchronous cameras were analyzed, and the time correction model of asynchronous cameras was established.In order to correct the time error of multiple asynchronous cameras, corresponding frame time synchronous matching was carried out by using the epipolar geometry constraints and the constraint principle of minimum distance from point to pole line.The shooting system composed of 2asynchronous cameras was built, the time matching error of corresponding frame was analyzed.The time error of start frame in video sequences was estimated, and the recovery of high precision dynamic scenes was realized.High-speed moving vehicle was simulated by using flashing LED, and the range of time error was estimated.In order to further validate the effectiveness of the method in actual road environment, four cameras were used to shoot the video sequences of freefall ball, thecomparative tests were respectively carried out by using time correction method based on similarity discrimination and the proposed method.Test result indicates that when the exposure time of camera is 1/30 s, and the cycle of high-speed flashing light with is 6.59 ms, the time error range of start frame is from 0to 6.59 ms, the corresponding frames ratio is above 88.1%.Compared with the time correction method based on similarity discrimination, the proposed method performs well.Compared with the several traditional time correction methods, the accuracies of time error matching and the 3Dreconstruction of asynchronous cameras in road monitoring system are significantly improved.
Keywords:intelligent transportation system  road monitoring system  machine vision  asynchronous cameras  time correction  epipolar geometry  corresponding frame matching
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《交通运输工程学报》浏览原始摘要信息
点击此处可从《交通运输工程学报》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号