利用自動車航行及多隻相機資料融合作室內環境虛擬實境模型之建構

Abstract

在虛擬實境的應用中，環境模組的建立通常需要花費相當的時間。如果我們欲建立一個和實際環境相仿的虛擬環境，在傳統的做法中，首先要用人工的方法測量出環境各部分的立體結構，爾後再建立出環境的三維模組。本論文發展了一套利用自動車航行來收集環境資訊，並以離線方式自動建立航行環境立體模組的方法。首先，由人開自動車航行於環境中，同時藉由裝置在自動車上的三隻相機，同步收集航行路徑上的環境資訊。然後藉由融合多隻相機所收集到的環境資訊建立出所航行環境的立體模組。我們還使用一些後處理來使建出的環境模組更完整，並且進一步將此模組以虛擬實境之方式呈現出來,讓使用者運用時有更強烈的參與感。

In applications of virtual reality, it is required, but often time consuming, to establish environment models. If we want to build a virtual environment based on a real one in terms of the traditional method, it is necessary to measure the stereo structure of each part of the environment. In this study, an offline automatic 3-D indoor virtual environment modeling method by the use of an autonomous land vehicle (ALV) navigation system is developed. An ALV is first driven manually along a path decided by a driver and the environment data are collected from three cameras mounted on the ALV at the same time. The system then generates the environment model by a data fusion approach using the model learning result from the multiple-view data, performs intelligent post-processing works to create a more complete model, and uses the model to construct a 3-D virtual reality (VR) model by the VRML language automatically. The result can be used for various applications in which users can have a better feeling of involvement in the VR environment.