Image-based 3D Object Modeling for Virtual Museum System

摘要

Virtual Reality technology provides a new methodology for human interface with realistic sensation, and has been received growing attention in the fields of human interface, computer graphics and computer vision. Reality of the viltual space is realized by generating photorealistic views in real time with computer graphic technology. One of key issues for the real time view generation problem is how to represent and display the 3D virtual space efficiently with adapting human vision sensitivity. In this paper we propose a novel and efficient approach for hierarchical modeling of 3D objects from multi-mode real images of range, color and many observing the objects, with which shape, texture, reflectance and internal structure are effectively described in multi-resolutions. Arbitrary views of the objects under arbitrary lighting conditions are then efficiently generated with sufficient accuracy by selecting the appropriate level of the model based on the distance between the observer and the object.