by Thomas Flanitzer
Abstract:
The eccentricity of a vertex is the longest shortest distance to any other vertex in a graph. We introduce the eccentricity transform which calculates the eccentricity for every point in a graph. Applied to digital images it offers some interesting properties including invariance to articulated motion and robustness whith respect to salt & pepper noise. Applied to graphs with an embedding it can be used for boundary determination. Its characteristics make it a good candidate for supporting or even replacing the distance transform as a basic tool in many feature extraction tasks (e.g. shape description). This report focuses on the computation of the eccentricity transform and explains implementation approaches.
Reference:
The eccentricity transform (computation) (Thomas Flanitzer), Technical report, PRIP, TU Wien, 2006.
Bibtex Entry:
@TechReport{TR107,
author = "Thomas Flanitzer",
title = "The eccentricity transform (computation)",
institution = "PRIP, TU Wien",
number = "PRIP-TR-107",
year = "2006",
url = "https://www.prip.tuwien.ac.at/pripfiles/trs/tr107.pdf",
abstract = "The eccentricity of a vertex is the longest shortest
distance to any other vertex in a graph. We introduce the eccentricity
transform which calculates the eccentricity for every point in a graph.
Applied to digital images it offers some interesting properties
including invariance to articulated motion and robustness whith
respect to salt \& pepper noise. Applied to graphs with an embedding
it can be used for boundary determination. Its characteristics make
it a good candidate for supporting or even replacing the distance
transform as a basic tool in many feature extraction tasks
(e.g. shape description). This report focuses on the computation
of the eccentricity transform and explains implementation approaches.",
}