Abstract
Automatic learning of 3D object descriptions for the recognition of future instances of sensed objects is an important feature of vision systems. In this paper, we describe a clustering-based method for learning object description and a scheme for subsequent classification using this description. The description derived is an evidence rulebase that discriminates between the object classes. During classification, a sensed object is either recognized as one of the objects in the database or rejected as an unknown object. The rules fired by the object are used to eliminate hypotheses for which there is no evidence in the sensed object and to arrive at its identity and pose.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
