节点文献

基于多传感器数据融合的路面三维重构

Study on the Road Three-Dimensional Reconstruction Based on Multi-Sensor Data Fusion

【作者】 黄波

【导师】 马荣贵;

【作者基本信息】 长安大学 , 计算机应用技术, 2010, 硕士

【摘要】 随着我国公路建设的发展,公路养护维修的多种任务也伴随而来,尤其是我国修建的一些高等级公路已进入中修或大修期。当前的三维检测技术已基本满足路面管理系统、智能交通系统(ITS)和日益完善的路面评价技术的需要,但落后的路面重构技术却无法将大量的检测数据应用到道路的养护维修中。目前,路面重构技术正向多传感器信息融合的方向发展。研究最新的数字路面重构技术不但可以满足我国日益繁重的路面养护维修的需要,而且对于提高我国的路面评价水平和ITS技术的发展也有着重要的意义。路面是一个三维实体,通过将路面三维实体映射到由路面里程、横断面和路面相对高程组成的三维路面坐标系中,可以将路面的三维重构简化为横断面重构、纵断面重构和高程重构。依据路面重构技术提出了路面的三层重构模型:纹理层,断面层和GPS层。由于每个层次具有不同的精度,反映不同的道路信息,因此为了还原真实路型,需要融合GPS信息(经纬度信息和高程信息);为了获取路面信息(车辙、平整度等),需要融合断面信息;为了得到路面细节(纹理,裂缝等),需要融合纹理信息。本文通过多信息融合实现了路面的可视化全频重构。通过对数字路面重构的准确性和完整性的研究,提出了GPS与惯导融合的连续定位和精密定位技术。通过对GPS误差的分析,得出引起误差最严重的因素是信号中断。本文利用惯性导航系统的特性,实现了在GPS信号中断时对道路的定位,保证了数据采集的完整性以及重构路面的完整性、连续性。传统的里程定位是通过光电编码器测距仪实现的,此方法在直道上精度很高。然而实际道路中弯道也占了很大一部分,由于内外道里程的差距,使得光电编码器测距仪在长距离定位中累积误差,定位精度越来越低。本文利用GPS和陀螺仪的辅助,解决了误差累积的问题,提高了里程定位的精度。传统的路面三维重构大多数仅反映了路面的部分信息,很难建立出立体直观的三维路面模型。真正的路面包括纹理、平整度、车辙、路型、高程、里程等信息。本文基于OpenGL生成了可视化的三维数字路面,为道路的养护维修提供了有效的参考。

【Abstract】 With the development of highway construction, road maintenance is also accompanied by a variety of tasks, in particular, some highways have entered a period of repair or overhaul. The three-dimensional measurement technology has basically met the requirements of Pavement Management System, Intelligent Transportation system (ITS) and increasingly sophisticated pavement evaluation. But, as the backward road reconstruction, a large number of detection data can not be applied to the roads maintenance and repair. At present, pavement reconstruction is developing to multi-sensor information fusion stage. We research on the latest digital pavement reconstruction technology, it not only can meet our increasing needs for road maintenance and repair, but also has great significance to improving road assessment level and promoting the development of ITS technology.Pavement is a three-dimensional body, under the three-dimensional road coordinate system with the profiling be x-axis, cross-sectional be y-axis and the relative height be z-axis, the three-dimensional reconstruction of road can be simplified to the cross-sectional reconstruction, vertical section reconstruction and height reconstruction. Based on the reconstruction technology, the three layers reconstruction model including texture layer, section layer and GPS layer is introduced. Each layer has the different accuracy and reflects the different road information, so the road needs to integrate GPS information(latitude, longitude and altitude) in order to restore the true road, integrate the section information to obtain the pavement information (rut, roughness, etc.) and integrate the texture information to get the road details (texture, cracks, etc.). In this paper, the full-range visualization reconstruction of road has been achieved by using multi-information fusion.By studying the accuracy and integrality of the pavement reconstruction, this paper gives a continuous and precision positioning technology with integration of GPS and INS. By analyzing the GPS error, this paper had made the conclusion that the most serious factor causing error is signal interruption. In this paper, achieved using the inertial navigation system features, though the GPS signal is interrupted, we can position on the road, the integrality of the data acquisition and the integrality and continuity of the pavement reconstruction have been ensured. The traditional road mileage positioning uses photoelectric encoder to achieve, this method is high precision in the straights. But the curve plays a major part in the reality road.Due to the difference of the mileage between inside lane and outside lane, the errors in the long-distance positioning have been accumulated and positioning accuracy is getting lower and lower. In this paper, we use GPS with gyroscope to solve the problem of error accumulation, increase the accuracy of the mileage positioning.The traditional three-dimensional pavement reconstruction reflects part of the road information, it is difficult to establish a intuitionistic three-dimensional model. The real road includes texture, smoothness, rut, road type, altitude, mileage, and more. In this paper, based on OpenGL, a three-dimensional visualization of digital road is generated, it provides the man who maintains road with a effective reference.

【关键词】 三维路面道路里程定位OpenGLGPS
【Key words】 Three-dimensional pavementmileage positioningOpenGLGPS
  • 【网络出版投稿人】 长安大学
  • 【网络出版年期】2011年 03期
  • 【分类号】TP391.41
  • 【被引频次】2
  • 【下载频次】167
节点文献中: 

本文链接的文献网络图示:

本文的引文网络