| Image-based rendering, IBR for short, is an area that gained in popularity during the last decade. There are two main objectives in this research: to provide a framework for IBR techniques and to apply IBR to dynamic scenes.;A novel taxonomy, called the rayset, is presented in this dissertation. The new taxonomy studies the scene representation techniques and the scene reconstruction techniques used in IBR separately. It is demonstrated that different image-based scene representations can all be cast as different kinds of raysets, and different scene reconstruction approaches can be regarded as attempts to render different raysets. Based on the taxonomy, existing IBR approaches are classified and the relationship between them is identified.;The rayset taxonomy also facilitates the development of new IBR techniques. Three novel IBR approaches, which are applications of the rayset concept in static environments, are presented. Each approach has different requirements and different advantages and can therefore be used for different applications.;Dynamic IBR research, which is still in its infancy, studies how to generate novel videos based on captured videos. One critical requirement for dynamic IBR is to reduce the sampling density. Based on the rayset concept, a new technique that can handle sparsely sampled scenes is discussed. A novel stereo vision algorithm is also presented, which provides accurate disparity information for the rendering process.;In summary, this dissertation provides several original contributions in the IBR area. Possible future research directions on how to employ temporal coherence in dynamic scenes are discussed. |
