A Theory of Origami World

The recovery of three-diniensional configurations o fa scene from its image is one of the most important stcps in coniputer vision. The Origami world is a model for understanding line drawings in terms of surfaces, and for finding their 3D ConfiRurafions. It assumes rho? surfaces themselves can be stand-alone objects, unlike the conventional rrihcdral world which assunies solid objects. We have established a labeling procedure for this Origami world, which canfind the 3-0 meaning of a given line drawing by assigninK one of the labels, + (convex edge), (concove edgc), t, and 4 (occluding boundary) to each line. The procedure uses afiltcring procedure not only for junction labels as in the Waltz labeling for the irihedral world, but also .for chccking the consistency of surface orientations. The theory includes the Hufnran-Clouv.c labeliti.q.c fiw the trihedrol solirl-object world as a subset. This paper also receals intcrestirig relationships anrong previous research in polyhedral scene analysis.