Recognition and pose determination of 3-D objects using multiple views

Abstract

We present a method for automatic recognition and pose (orientation) determination of 3-D objects of arbitrary shape. The approach consists of an off-line stage in which the recognition and pose identification plan is derived and an on-line recognition and pose identification stage. To obtain the plan, the objects are observed from all possible views and for each view a shape feature vector is extracted. These vectors are then used to structure the views by a binary decision tree. Associated with each node in the decision tree is a measure indicating the reliability of making a correct decision at that particular node. This measure drives the procedure for an optimal next-view planning when additional views are necessary to resolve the ambiguities. The output of the first stage is the recognition-pose-identification plan which then guides the recognition and pose determination of an unknown object in an unknown pose. The system has been tested on a set of real objects using multiresolution NIP (non-information-preserving) shape features to characterize the views.