CSS-AFFN: A Dataset Representation Model for Active Recognition Systems

摘要

This paper proposes an object database representation model for active recognition systems. This model optimizes dataset information. The objects are modeled by using the proposed Canonical Sphere Section (CSS) model and the shape is normalized to affine transformations in the spectral dominium. This dataset representation model is compared with other shape representation models and implemented in an active recognition system which develops object manipulation. Its feasibility in complex robotic applications is therefore proved.