Photo-based automatic 3D reconstruction of train accident scenes.

Tang, Z., Nie, Y., Chang, J., Zhang, J. J. and Liu, F., 2018. Photo-based automatic 3D reconstruction of train accident scenes. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 232 (1), 144 - 158.

Full text available as:

[img]
Preview
PDF
Photo-based automatic 3D reconstruction of train accident scenes_submitted.pdf - Submitted Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

1MB

DOI: 10.1177/0954409716662089

Abstract

Railway accidents place significant demands on the resources of, and support from, railway emergency management departments. Once an accident occurs, an efficient incident rescue plan needs to be delivered as early as possible to minimise the loss of life and property. However, in the railway sector, most relevant departments currently face a challenge in drawing up a rescue scheme effectively and accurately with the insufficient information collected from the scene of a train accident. To assist with the rescue planning, we propose a framework which can rapidly and automatically construct a 3D virtual scene of a train accident by utilising photos of the accident spot. The framework uses a hybrid 3D reconstruction method to extract the position and pose information of the carriages involved in an accident. It adopts a geographic information system and a 3D visualisation engine to model and display the landscapes and buildings at the site of a train accident. In order to assess and validate our prototype, we quantitatively evaluate our main algorithm and demonstrate the usage of our technology with two case studies including a simulated scene with an in-lab setting and a real train derailment scene from on-site pictures. The results of both are accoun table with high accuracy and represent the ability of timely modelling and visualisation of a train accident scene.

Item Type:Article
ISSN:0954-4097
Uncontrolled Keywords:Train accident; scene reconstruction; accident rescue; photo-based modelling; railway safety
Group:Faculty of Media & Communication
ID Code:30290
Deposited By: Unnamed user with email symplectic@symplectic
Deposited On:31 Jan 2018 15:15
Last Modified:31 Jan 2018 15:15

Downloads

Downloads per month over past year

More statistics for this item...
Repository Staff Only -