ABSTRACT
This chapter shows how the D ichrom atic Reflection Model and the seg m entation results of C hapter 4 can be used to generate intrinsic body and surface reflection images of the scene. The body reflection im age shows the scene w ithout highlights, while the surface reflection im age shows only the highlights. Since the reflection images have a sim pler relationship to the illum ination geom etry th an the original im age, they m ay improve the results of m any other com puter vision algorithm s, such as m otion analysis, stereo vision, and shape from shading or highlights.
