This paper describes a process by which wire frames of polyhedra may be generated from multiple range views of a scene. Line segments are extracted from each view and transformed into a world coordinate system to produce the wire frames. These wire frames are used in the construction of full polyhedral models with the algorithm of Markowsky and Wesley.

This paper presents a method to evaluate surface curvature at points on the three-dimensional shell of an object using a volume representation. By considering points where the curvature is significantly high to be feature points, curves may be constructed on the surface of the object model. Such descriptive curves will become useful in finding orientation or making identification in robotics applications. Explicit formulation and experimental results of curvature operators and a corner-finding operator are presented.

Vision will be increasingly important in robotics applications. An object is modeled by critical curves extracted from the object. In the simplest case the object's curve is its outline. A solution to the curve partitioning problem is shown for nonconvex objects. Curves are described in a rotation and translation invariant way. A way to build a combined model database of many classes of objects is presented. A test to insure disjointness of model classes is given. An efficient technique for computing a network of these curves from a gray-level frame is presented. A graph algorithm is presented to match the model in an efficient way, independent of the scaling found in the scene. A technique for computing and classifying more general critical curves from three-dimensional data is developed.