Study on a method to precisely reconstruct real environments using metrical techniques and seamlessly integrate a mixed real/virtual world.

点击翻译按钮获取中文摘要

In this study, we have explored a new approach to reconstruct 3D geometric shapes of real outdoor and elastic objects with a high degree of accuracy. While traditional approaches cannot treat these objects, we have developed a new method to overcome the problem. The developed method integrate metrical techniques and 3D measurement technologies based on laser scanning and vision. It flexibly allows to reconstruct, visualize, and deform the objects with variable precision for 3D reconstruction. We further developed some application systems to evaluate effectiveness of the developed methods. This study has produced the following results :1.The 3D shape reconstruction method using Fourier transform is designed and implemented. It converts a point-could data acquired from a 3D range finder into a structured 3D data format to accurately reconstruct and flexibly deform the measured objects in a virtual environment.2.The 3D geometric modeling method of the reconstructed objects is invented. It allows users to intuitively deform the reconstructed 3D objects through a freehand sketch interface and an interactive genetic algorithm-based shape exploration strategy.3.The 3D modeling application systems are developed. Three application systems for 3D texture authoring, web icon design, and 3D indexing and retrieving of ceramics are implemented to evaluate the devised method.4.The experiments of the networked virtual environment are conducted. We have implement a network experiment test bed to assess and verify the effectiveness of the 3D data sharing and remote collaboration using the developed method. We have also implemented a virtual urban planning system to evaluate how the developed method can treat the real-world objects such as buildings and houses.These results are published in IEEE/ACM conference papers and IPSJ/IEICE journal papers.