Image-Based Rendering and the Sampling of the Plenoptic Function

Image-based rendering (IBR) is a technique for producing arbitrary views of a scene using multiple images instead of exact object models. The central concept is that each image comprises a collection of light rays and a new view is interpolated from these light rays. If we modelled the light rays using a seven-dimensional function, known as the plenoptic function, then IBR can be viewed in terms of sampling and reconstruction. Therefore the important goal of minimizing the number of images required in IBR, whilst maintaining rendering quality, can be examined through sampling analysis of the plenoptic function. In this context, the chapter examines the state of the art in plenoptic sampling theory. It focuses on both uniform and adaptive sampling of the plenoptic function. In particular, it presents theoretical results for uniform sampling based on spectral analysis of the plenoptic function and algorithms for adaptive plenoptic sampling.