基于单目图像的列车事故场景三维重建

发布时间：2018-06-01 13:29

本文选题：轨道车辆 + 事故救援　；参考：《交通运输工程学报》2017年01期

【摘要】：为了辅助铁路事故应急救援方案的制定,提出了一种基于单目图像的列车事故场景快速三维重建方法。考虑不同应用场景的2种相机投影模型,采用SIFT算法提取图像特征并与事故列车的CAD模型相匹配,通过引入车厢之间的几何约束,将三维重建转换为求解带约束的非线性最小二乘问题,最终还原事故主体的位置与姿态。为了定量与定性验证该方法的计算效果,分别对模拟列车事故场景与真实列车事故场景进行车厢的三维重建。在模拟列车事故场景中采用了较精确的有限相机投影模型进行离线标定,在真实列车事故场景中采用了较稳定的针孔模型进行自标定。分析结果表明:通过对模拟场景的定量分析,重建两节车厢中用于测量的8个节点的最大相对误差为4.54%,平均相对误差为1.85%;通过对真实场景的定性分析,结合地形信息校正,同样能够实现车厢位置与姿态的三维还原;最终借助三维可视化引擎,可在视觉上还原整个事故环境全貌。该方法还可用于应急救援电子沙盘的开发以进行铁路事故分析和安全教育。
[Abstract]:A fast 3D reconstruction method of train accident scene based on monocular image was proposed to assist the formulation of emergency rescue scheme for railway accident. Considering two kinds of camera projection models in different application scenarios, the SIFT algorithm is used to extract the image features and match with the CAD model of the accident train, and the geometric constraints between the cars are introduced. The 3D reconstruction is transformed into solving the nonlinear least square problem with band constraints, and finally the position and attitude of the subject are restored. In order to quantitatively and qualitatively verify the effect of this method, the simulated train accident scene and the real train accident scene were reconstructed respectively. In the simulation of train accident scene, a more accurate finite camera projection model is used for off-line calibration, and a more stable pinhole model is used for self-calibration in the real train accident scene. The results show that the maximum relative error is 4.54 and the average relative error is 1.85 through the quantitative analysis of the simulated scene, and the correction of terrain information is combined with the qualitative analysis of the real scene. It can also realize the three-dimensional restoration of the position and attitude of the carriage, and finally, with the help of the three-dimensional visualization engine, the whole scene of the accident environment can be visually restored. The method can also be used in the development of emergency rescue electronic sand table for railway accident analysis and safety education.
【作者单位】：西南交通大学牵引动力国家重点实验室;伯恩茅斯大学英国国家计算机动画中心;
【基金】：国家自然科学基金项目(51405402,51475394) 牵引动力国家重点实验室自主研究课题(2015TPL_T06) 中央高校基本科研业务费专项资金项目(2682016CX128)
【分类号】：U298.5;TP391.41

【相似文献】