RECOGNIZING POLYHEDRAL OBJECTS FROM A SINGLE PERSPECTIVE VIEW

This paper considers the problem of recognizing 3D polyhedral objects from a single perspective image. A hypothesis-verification paradigm ba sed on a local shape representation is presented. The proposed system is designed to recognize polyhedral objects containing at least one co mpletely visible spatial edge bounded by two vertices of the same obje ct. In the framework, 2D vertices interpreted as the projection of a t rihedral vertex which is a 3D spatial vertex with three lines emanatin g from the tip are employed as seed features for model invocation and hypothesis generation. To simplify the perspective analysis, Kanatani1 has proposed an intuitive and elegant technique. Using the technique, we derive a fourth-degree polynomial for interpreting a trihedral ver tex. The contribution of our solution is that there are no restriction s on angles between the vertex edges. To reduce the number of hypothes es generated from scene-model vertex assignments, and recover the comp lete object pose, we propose a composite feature, namely vertex-CS fea ture by combining a trihedral vertex and a V-junction which share a co mmon edge. The geometric constraint of this composite feature is deriv ed. A matching strategy used in the recognition system is discussed. T he feasibility of the proposed method is illustrated on real data.