Soft Image Segmentation: On the Clustering of Irregular, Weighted, Multivariate Marked Networks

Details

Ressource 1Download: Cere_Bavaud_SoftImageSegm_2019.pdf (615.32 [Ko])
State: Public
Version: Author's accepted manuscript
Serval ID
serval:BIB_2BA2239595E3
Type
A part of a book
Publication sub-type
Chapter: chapter ou part
Collection
Publications
Institution
Title
Soft Image Segmentation: On the Clustering of Irregular, Weighted, Multivariate Marked Networks
Title of the book
Communications in Computer and Information Science
Author(s)
Ceré Raphaël, Bavaud François
Publisher
Springer International Publishing
ISBN
9783030060091
9783030060107
ISSN
1865-0929
1865-0937
Publication state
Published
Issued date
2019
Peer-reviewed
Oui
Editor
Ragia   L., Laurini R., Rocha  J.
Series
CCIS vol 936
Genre
Geographical Information Systems Theory, Applications and Management
Pages
85-109
Language
english
Notes
(revised Selected Papers of GISTAM 2017)
Abstract
The contribution exposes and illustrates a general, flexible formalism, together with an associated iterative procedure, aimed at determining soft memberships of marked nodes in a weighted network. Gathering together spatial entities which are both spatially close and similar regarding their features is an issue relevant in image segmentation, spatial clustering, and data analysis in general. Unoriented weighted networks are specified by an ``exchange matrix", determining the probability to select a pair of neighbors. We present a family of membership-dependent free energies, whose local minimization specifies soft clusterings. The free energy additively combines a mutual information, as well as various energy terms, concave or convex in the memberships: within-group inertia, generalized cuts (extending weighted Ncut and modularity), and membership discontinuities (generalizing Dirichlet forms). The framework is closely related to discrete Markov models, random walks, label propagation and spatial autocorrelation (Moran's I), and can express the Mumford-Shah approach. Four small datasets illustrate the theory.
Keywords
free energy, image segmentation, iterative clustering, soft K-means, Laplacian, modularity, Moran's I, Mumford-Shah functional, multivariate features, Ncut, soft membership, spatial autocorrelation, spatial clustering
Open Access
Yes
Create date
28/01/2019 19:13
Last modification date
20/08/2019 13:11
Usage data