RUS  ENG
Full version
JOURNALS // Computer Optics // Archive

Computer Optics, 2014 Volume 38, Issue 3, Pages 511–520 (Mi co228)

IMAGE PROCESSING, PATTERN RECOGNITION

Morphological shape descriptors of binary images based on elliptical structuring elements

S. V. Sidyakin, Yu. V. Vizilter

FGUP “GosNIIAS”

Abstract: In this paper, we proposed algorithms for constructing continuous skeletons with elliptical structuring element (SE). The transformation between disk and elliptical skeletons is described. Computationally efficient discrete-continuous approach for the construction of morphological descriptors (spectra and maps) with fixed elliptical SE is proposed based on elliptical skeletons. The definitions of morphological elliptical maps, size spectra, directions spectra, elongation spectra with arbitrary elliptical structuring element is proposed. Definitions of two-dimensional spectra with various combinations of size and shape factors is given. Proposed morphological descriptors can be used directly for shape comparison or for shape segmentation into simple geometric parts of specified thickness, direction and elongation.

Keywords: mathematical morphology, pattern spectra, continuous skeleton, ellipse.

Received: 17.06.2014



© Steklov Math. Inst. of RAS, 2024