A Graph-Theoretic Method for Decomposing Two-Dimensional Polygonal Shapes into Meaningful Parts

Authors:
C. Guerra;G. G. Pieroni
Affiliations:
Dipartimento di Matematica, Università/ della Calabria Arcavacata, Italy/ Istituto di Automatica, Università/ di Roma, Roma, Italy.;Dipartimento di Matematica, Università/ della Calabria, Arcavacata, Italy.
Venue:
IEEE Transactions on Pattern Analysis and Machine Intelligence
Year:
1982

Citing 0
Cited 2

A method for detecting correspondences in a sequence of modifying shapes

Pattern Recognition Letters
On a conceptual description of images

Pattern Recognition Letters

Quantified Score

Hi-index	0.14

Visualization

Abstract

This note gives a new graph-theoretic shape decomposition procedure, extending the idea in the paper ``Decomposition of Two-Dimensional Shapes by Graph-Theoretic Clustering'' by Shapiro and Haralick [1]. A binary matrix is used in order to express the LI relation. A solution for the problems connected with artificial separation of two CS cliques and the presence of a regular concavity is proposed. Our algorithm produces nonoverlapping shape parts in much reduced execution time.