Abstract
The generation of virtual environments has recently attracted much attention. In this paper we address our recent work in this area. Unlike most of methods of 3D scene construction, instead of pursuing the precise 3D structure, we consider the 3D pseudo-structure of 3D scenes to construct the virtual environments from multi-images with forward motion. For each frame, visual 3D geometry of the scene's background is defined as a simple polygonal model, by roughly specifying the vanishing point. Modeled 3D scene is decomposed into sub-images, which are corresponding to the polygons in the modeled 3D scene. Sub-images whose polygons have the same normal are registered using the Levenberg-Marquardt iterative algorithm to estimate motion parameters. These registered sub-images are correspondingly pasted together to form a pseudo-3D space. By controlling the position and direction, the virtual camera can walk through in this virtual space to create novel 2D views and to acquire an immersive impression. Even if the virtual camera goes deeper into this virtual environment, it also can obtain a novel view while maintaining high resolution.
