Abstract
Morphological operations can be efficiently computed on parallel architectures using the decomposition of the structuring elements. In some cases, decomposition is guided to optimize computation for a given underlying hardware, but in other cases it is the shape of the structuring elements which directs the decomposition. In this paper we present a method to decompose disks. Morphological operations with isotropic structuring elements present interesting properties as shape and size descriptors. The method developed is based on a constraint-satisfaction algorithm that gives an optimal decomposable disk. Optimality is given by the shape of the disk since it is the best discrete approximation of a circle that allows a 3 × 3 decomposition. © 1997 Elsevier Science B.V.
Original language | English |
---|---|
Pages (from-to) | 845-854 |
Journal | Image and Vision Computing |
Volume | 15 |
Issue number | 11 |
Publication status | Published - 1 Jan 1997 |
Keywords
- Discrete disk approximation
- Disk decomposition
- Mathematical morphology
- Structuring element