1st Canadian Conference on Computer and Robot Vision (CRV'04)
Recovering the Shading Image under Known Illumination
University of Western Ontario, London, Ontario, Canada
May 17-May 19
ISBN: 0-7695-2127-4
Given a standard RGB color image, an approach to recovering the shading image, an intrinsic image that depends only on the geometry of the viewing situation, is presented. The approach is formulated for scenes that exhibit both diffuse and specular properties. It is discriminative in the sense that variations in intensity caused by material changes are .rst detected and then removed. Material changes are detected using a method based on color information alone. The method stems from a scheme proposed here to handle specularities under known scene illumination chromaticity. This scheme is also used to produce an image that is free of specular re.ections. Any material changes are then removed from this specular-free image by using an inpainting technique. The performance of the proposed approach has been tested on synthetic images.
Index Terms:
2D/3D Scene Analysis, Shape Analysis.
Citation:
Javier Toro, Djemel Ziou, Marie-Flavie Auclair-Fortier, "Recovering the Shading Image under Known Illumination," crv, pp.92-96, 1st Canadian Conference on Computer and Robot Vision (CRV'04), 2004