Inference of 3-D shape from 2-D contours in a single image is an important problem in machine vision. Often, techniques to solve this problem examine each surface in the scene separately whereas our perception of their shapes clearly depends on the interplay between them as well. In this paper, we describe a technique that attempts to recover the shapes of all the surfaces of an object simultaneously, though it is limited to objects made of zero-Gaussian curvature surfaces. Our technique is based on an analysis of three kinds of symmetries defined in the paper and the constraints that derive from them, and from other boundaries. This technique uses some of the constraints developed in an earlier paper that was limited to examining a zero-Gaussian curvature surface cut by parallel planes. This restriction has been removed here and the constraints have been reformulated to allow integration of constraints from all the neighboring surfaces. Results on some complex examples are shown. © 1994.