Interreflections
Shadows and interreflections are present in all real scenes and provide a rich set of cues for vision.  One interesting fact about shadows and interreflections is that they are intrinsically related.  Shadows tend to occur in those parts of a scene in which interreflections can have the largest effect.  In a theoretical paper in IJCV 1999, I provided several basic  results concerning this relationship in terms of the Koenderink and van Doorn's "interreflection modes" of a scene.  I showed that for a given scene, the  interreflection mode having the largest gain is a physically realizable radiance function i.e. it is non-negative.  I also derived bounds on the gain of this mode, discussed how this mode is related to shadows, and analyzed how well an n-bounce model of interreflections approximates an infinite-bounce model and how shadows affect this  approximation.  I also introduced a novel method for inferring  surface color in a uni-chromatic scene.  The method is based on the relative contrast of the scene in different color channels.

The idea about color is related to work that Alan Gilchrist and colleagues have done, in which they had human observers look into a small room and make judgments about surface reflectance i.e. whether the surfaces are white, light grey, dark grey, black, etc.    The idea is that interreflections can provide a cue to lighting, since a white painted room tends to have much more indirect light than a dark painted room.    To illustrate this effect, consider the image below on the left which was generated using computer graphics (a package called RADIANCE).     We can decompose this image into its direct component (due to the light source only) and its indirect component (which is due only to interreflected light).   Notice that the image on the right is not constant intensity.   

Selected Publications
image    =     image   +    image