Development of Driving Simulation System based on Real Image Processing and its Evaluation on Visual Characteristic

点击翻译按钮获取中文摘要

In this research, a method to construct photo-realistic driving view for a driving simulator (DS) is developed. As for bacgrounds of a DS, real videos captured by our sensing vehicle are reconstructed, and displayed with free-point view and data compression. As for foregrounds, CG of road structures became enabled to be quickly modeled from national digital maps. Also, the visual characteristic of rendering methods for a DS is evaluated by comparing two types, one is free-point-view rendering based on scene depth, and the other is a simple rendering based on a cylindrical depth map that can only deal with rotation component of the motion of the DS. As a result, it was implied that there is a possibility that simple rendering method can be superior for rendering complex traffic scenes.