Image-based rendering techniques for application in virtual environments.

Abstract

In this thesis, the methods of Image-Based Rendering for creating virtual environment applications are studied in order to construct a real image-based virtual environment with the principle of representing an environment through a set of pre-captured images. These pre-captured images are used to synthesize arbitrary views in a virtual environment. Currently developed techniques include view synthesis through interpolation, Light Field Rendering and Concentric Mosaics. These methods are presented and compared in the thesis and we conclude that the Concentric Mosaics technique is more suitable for practical applications. The stereoscopic view synthesis through the Concentric Mosaics rendering technique is also addressed. The application of the anaglyph technique can make stereo views of a virtual environment available to any personal computer user with an inexpensive pair of colored glasses.