Authors:
Raphaël Ceré
and
François Bavaud
Affiliation:
University of Lausanne, Switzerland
Keyword(s):
Free Energy, Image Segmentation, Iterative Clustering, K-means, Laplacian, Modularity, Multivariate Features, Ncut, Soft Membership, Spatial Autocorrelation, Spatial Clustering.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Business Analytics
;
Cardiovascular Technologies
;
Computational Geometry
;
Computer Vision, Visualization and Computer Graphics
;
Computing and Telecommunications in Cardiology
;
Data Engineering
;
Decision Support Systems
;
Decision Support Systems, Remote Data Analysis
;
Health Engineering and Technology Applications
;
Image Formation and Preprocessing
;
Knowledge-Based Systems
;
Symbolic Systems
Abstract:
Image segmentation and spatial clustering both face the same primary problem, namely to gather together spatial entities which are both spatially close and similar regarding their features. The parallelism is particularly obvious in the case of irregular, weighted networks, where methods borrowed from spatial analysis and general data analysis (soft K-means) may serve at segmenting images, as illustrated on four examples. Our approach considers soft memberships (fuzzy clustering) and attempts to minimize a free energy functional made of three ingredients : a within-cluster features dispersion (hard K-means), a network partitioning objective (such as the Ncut or the modularity) and a regularizing entropic term, enabling an iterative computation of the locally optimal soft clusters. In particular, the second functional enjoys many possible formulations, arguably helpful in unifying various conceptualizations of space through the probabilistic selection of pairs of neighbours, as well a
s their relation to spatial autocorrelation (Moran's I).
(More)