Toward efficient morphological shape representation

Because mathematical morphology is well-suited to capturing geometric information, morphology-based approaches have been popular for shape representation. The morphological skeleton and recently proposed morphological shape decomposition (MSD) are the primary competing morphology-based approaches for shape representation. Unfortunately, little has been done for these schemes to reduce representation redundancy and exploit tolerable representation error. We propose a shape representation scheme that is typically more computationally efficient than the morphological skeleton and MSD. Our method greatly augments the MSD. We introduce new constituent component types and incorporate a cost-based search strategy for finding an efficient representation. Further, if some representation error is tolerable, our technique can give still more efficient representations.