Image-Based 3D Scene Reconstruction and Rescue Simulation Framework for Railway Accidents

Abstract

Although the railway transport is regarded as a relatively safe transportation tool, many railway accidents have still happened worldwide. In this research, an image-based 3D scene reconstruction framework was proposed to help railway accident emergency rescues. Based on the improved constrained non-linear least square optimization, the framework can automatically model the accident scene with only one panorama in a short time. We embedded the self-developed global terrain module into the commercial visualization and physics engine, which makes the commercial engine can be used to render the static scene at anywhere and simulate the dynamic rescue process respectively. In addition, a Head Mounted Device (HMD) was integrated into this framework to allow users to verify their rescue plan and review previous railway accidents in an immersive environment.