Goniometric Light Reconstruction for Augmented Reality Image Synthesis
Thorsten Grosch · 1997
Abstract. Reconstruction of light and material properties from digital photographs is a challenging problem. We describe a new image-based approach for the reconstruction of light and material properties from a set of photographs. Using these values in a lighting simulation makes it possible to augment the photograph with correct lighting conditions in all regions of the photograph. Since light sources are the most important component for the correct appearance of virtual objects, special emphasis is put on the reconstruction of a light goniometric as a realistic light source description. Furthermore we describe a new data structure to rapidly detect changing regions in the photograph for a fast update after moving objects. For high quality images we show how to use anti-aliasing techniques to improve the display quality Inserting virtual objects in photographs with correct appearance is an active area of research in the last few years. The first attempt was made by Nakamae et al.[Na86], where virtual buildings were placed in a photograph. The base for most of the following work was done by Alain Fournier et al.[FGR93]. They introduced the differential rendering