Viewpoint Quality: Measures and Applications

Abstract

Several methods that use the notion of viewpoint quality have been recently introduced in different areas of computer graphics, such as scene understanding, exploration of virtual worlds, radiosity and global illumination, image-based rendering and modelling. In this paper, we analyze the behavior of three different viewpoint quality measures. The first one is a heuristic measure, the second one is the viewpoint entropy, and the third one is a new measure based on the Kullback-Leibler distance between the projected and actual distributions of the areas of the polygons in the scene. In addition, this paper reviews different applications and introduces a new algorithm using the Kullback-Leibler distance for the selection of a representative set of n views. Our method is based in selecting the view that minimizes the Kullback-Leibler distance between the mixture of the distributions of all selected views and the actual area distribution.