Professional Documents
Culture Documents
Textbook Image Analysis and Recognition 11Th International Conference Iciar 2014 Vilamoura Portugal October 22 24 2014 Proceedings Part I 1St Edition Aurelio Campilho Ebook All Chapter PDF
Textbook Image Analysis and Recognition 11Th International Conference Iciar 2014 Vilamoura Portugal October 22 24 2014 Proceedings Part I 1St Edition Aurelio Campilho Ebook All Chapter PDF
https://textbookfull.com/product/image-analysis-and-
recognition-11th-international-conference-iciar-2014-vilamoura-
portugal-october-22-24-2014-proceedings-part-ii-1st-edition-
aurelio-campilho/
https://textbookfull.com/product/image-analysis-and-
recognition-17th-international-conference-iciar-2020-povoa-de-
varzim-portugal-june-24-26-2020-proceedings-part-i-aurelio-
campilho/
https://textbookfull.com/product/image-analysis-and-
recognition-17th-international-conference-iciar-2020-povoa-de-
varzim-portugal-june-24-26-2020-proceedings-part-ii-aurelio-
campilho/
https://textbookfull.com/product/computational-intelligence-
international-joint-conference-ijcci-2014-rome-italy-
october-22-24-2014-revised-selected-papers-1st-edition-juan-
julian-merelo/
Image Analysis
and Recognition
11th International Conference, ICIAR 2014
Vilamoura, Portugal, October 22–24, 2014
Proceedings, Part I
123
Lecture Notes in Computer Science 8814
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Alfred Kobsa
University of California, Irvine, CA, USA
Friedemann Mattern
ETH Zurich, Zürich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
Oscar Nierstrasz
University of Bern, Bern, Switzerland
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbruecken, Germany
More information about this series at http://www.springer.com/series/7412
Aurélio Campilho Mohamed Kamel (Eds.)
•
Image Analysis
and Recognition
11th International Conference, ICIAR 2014
Vilamoura, Portugal, October 22–24, 2014
Proceedings, Part I
123
Editors
Aurélio Campilho Mohamed Kamel
Faculty of Engineering Department of Electrical and Computer
University of Porto Engineering
Porto University of Waterloo
Portugal Waterloo, ON
Canada
LNCS Sublibrary: SL6 – Image Processing, Computer Vision, Pattern Recognition, and Graphics
This is the 11th edition of the ICIAR series of annual conferences offering an
opportunity for the participants to interact and present their latest research in theory,
methodology, and applications of image analysis and recognition. ICIAR 2014, the
International Conference on Image Analysis and Recognition, was held in Vila Moura,
Portugal, October 22–24, 2014. ICIAR is organized by AIMI – Association for Image
and Machine Intelligence, a not-for-profit organization registered in Ontario, Canada.
For ICIAR 2014, we received a total of 177 full papers from 39 countries. Before
the review process all the papers were checked for similarity using a comparison
database of scholarly work. The review process was carried out by members of the
Program Committee and other reviewers. Each paper was reviewed by at least two
reviewers, and checked by the conference chairs. A total of 107 papers were finally
accepted and appear in the two volumes of this proceedings. We would like to sincerely
thank the authors for responding to our call, and we thank the reviewers for the careful
evaluation and feedback provided to the authors. It is this collective effort that resulted
in the strong conference program and high-quality proceedings.
Each year we attempt to focus on a specific topic for the keynote speeches and
conduct a panel discussion on the topic.
This year, the conference theme was focused on the topic “Sparse Representations
for Image Analysis and Recognition.” We were very pleased to include three out-
standing keynote talks on this topic: “Optimization Algorithms for Sparse Represen-
tations: Some History and Recent Developments” by Mário Figueiredo, Instituto
Superior Técnico Portugal; “Morphological Diversities in Astrophysics” by Jean-Luc
Starck, CosmoStat Laboratory, France; and “Sparse Stochastic Processes with Appli-
cation to Biomedical Imaging” by Michael Unser, Ecole Polytechnique Fédérale de
Lausanne, Switzerland. The keynote speakers also participated in the panel “Sparse
Representation for Image Analysis and Recognition: Trends and Applications.” We
would like to express our gratitude to the keynote speakers for accepting our invitation
to share their vision and recent advances in their areas of expertise, which are at the
core of the topics of the conference.
We would like to thank Khaled Hammouda, the webmaster of the conference, for
maintaining the Web pages, interacting with the authors, and preparing the
proceedings.
As all conferences, the success of ICIAR 2014 is attributed to the effort and work of
many people, including members of the Organizing Committee, staff, and volunteers.
We gratefully acknowledge their support and efforts.
We are also grateful to Springer’s editorial staff for supporting this publication in the
LNCS series. We also would like to acknowledge the professional service of Viagens
Abreu in taking care of the registration process and the special events of the
conference.
VI Preface
Finally, we are very pleased to welcome all the participants to ICIAR 2014. For
those who were not able to attend, we hope this publication provides a good view into
the research presented at the conference, and we look forward to meeting you at the
next ICIAR conference.
General Chairs
Aurélio Campilho University of Porto, Portugal
Mohamed Kamel University of Waterloo, Canada
Conference Secretariat
Viagens Abreu SA, Portugal
Webmaster
Khaled Hammouda Waterloo, Ontario, Canada
Advisory Committee
M. Ahmadi University of Windsor, Canada
P. Bhattacharya Concordia University, Canada
T.D. Bui Concordia University, Canada
M. Cheriet University of Quebec, Canada
E. Dubois University of Ottawa, Canada
Z. Duric George Mason University, USA
G. Granlund Linköping University, Sweden
L. Guan Ryerson University, Canada
M. Haindl Institute of Information Theory and Automation,
Czech Republic
E. Hancock University of York, UK
J. Kovacevic Carnegie Mellon University, USA
M. Kunt Swiss Federal Institute of Technology (EPFL),
Switzerland
VIII ICIAR 2014
Program Committee
A. Abate University of Salerno, Italy
M. Ahmed Wilfrid Laurier University, Canada
L. Alexandre University of Beira Interior, Portugal
J. Alirezaie Ryerson University, Canada
G. Andreu-Garcia Universitat Politècnica de València, Spain
H. Araújo University of Coimbra, Portugal
Emilio Balaguer-Ballester Bournemouth University, UK
T. Barata University of Coimbra, Portugal
J. Barbosa University of Porto, Portugal
J. Batista University of Coimbra, Portugal
R. Bernardes University of Coimbra, Portugal
A. Bezerianos National University of Singapore, Singapore
J. Bioucas Technical University of Lisbon, Portugal
I. Bloch Télécom ParisTech, France
T.D. Bui Concordia University, Canada
C. Busch Gjøvik University College, Norway
F. Camastra University of Naples Parthenope, Italy
J. Cardoso University of Porto, Portugal
G. Carneiro University of Adelaide, Australia
M. Coimbra University of Porto, Portugal
M. Correia University of Porto, Portugal
J. Debayle Ecole Nationale Supérieure des Mines
de Saint-Étienne, France
J. Dias University of Coimbra, Portugal
G. Doretto West Virginia University, USA
H. du Buf University of the Algarve, Portugal
J. Fernandez Centro Nacional de Biotecnología – CSIC, Spain
I. Fondón University of Seville, Spain
A. Fred Technical University of Lisbon, Portugal
G. Freeman University of Waterloo, Canada
ICIAR 2014 IX
Reviewers
Supported by
Sparse Representation
Remote Sensing
Applications
Biometrics
Periodic Background Pattern Detection and Removal for Cell Tracking . . . . 123
Tiago Esteves, ¢ngela Carvalho, Fernando Jorge Monteiro,
and Pedro Quelhas
Nerve Detection in Ultrasound Images Using Median Gabor Binary Pattern. . . 132
Oussama Hadjerci, Adel Hafiane, Pascal Makris, Donatello Conte,
Pierre Vieyres, and Alain Delbos
Computer-Aided Diagnosis
Grading Cancer from Liver Histology Images Using Inter and Intra Region
Spatial Relations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Micka˝l Garnier, Maya Alsheh Ali, Johanne Seguin, Nathalie Mignet,
Thomas Hurtut, and Laurent Wendling
3D Imaging
Adaptive Feature Selection for Object Tracking with Particle Filter . . . . . . . 395
Darshan Venkatrayappa, D¯sir¯ Sidib¯, Fabrice Meriaudeau,
and Philippe Montesinos
Robot Vision
Multiple Camera Approach for SLAM Based Ultrasonic Tank Roof Inspection . . . 453
Christian Freye, Christian Bendicks, Erik Lilienblum,
and Ayoub Al-Hamadi
Abstract. Graph and tree-like structures such as blood vessels and neu-
ronal networks are abundant in medical imaging. We present a method
to calculate path descriptors in geometrical graphs, so that the similar-
ity between paths in the graphs can be determined efficiently. We show
experimentally that our descriptors are more discriminative than exist-
ing alternatives. We further describe how to match two geometric graphs
using our path descriptors. Our main application is registering images
for which standard techniques are inefficient, because the appearance
of the images is too different, or there is not enough texture and no
uniquely identifiable keypoints to be found. We show that our approach
can register these images with better accuracy than previous methods.
1 Introduction
Blood vessels, nerve fibers or pulmonary airways are examples of biological struc-
tures that can be represented as geometrical graphs with nodes corresponding
to branching points and edges corresponding to curves connecting the branching
points (Fig. 1). We consider the problem of registering two 2D or 3D images
based on a common geometric graph structure both images contain and which
we assume to be already extracted (e.g. [1]). This approach has the potential of
being much faster than standard pixel-based image registration techniques [2]
and tolerate very different image appearances. With respect to key point regis-
tration methods [3], registering geometric graphs provides more clues.
Unlike most existing approaches, our method can in a reasonable time handle
rather general transformations and large displacements, as well as partial over-
laps. The key contribution over our previous work [4,5] is an alternative coarse
alignment step based on finding similarities between path descriptors to restrict
the set of possible correspondences and thus make the matching more efficient.
2 Related Work
Ignoring edges, the geometric graph registration becomes a point cloud matching
problem, which can be solved by RANSAC-like approaches [6,7] These methods
c Springer International Publishing Switzerland 2014
A. Campilho and M. Kamel (Eds.): ICIAR 2014, Part I, LNCS 8814, pp. 3–11, 2014.
DOI: 10.1007/978-3-319-11758-4 1
4 M.A. Pinheiro and J. Kybic
do not need initialization but only work for restricted class of transformations,
such as rigid or affine. On the other hand ICP-like approaches [8,9] can handle
nonlinear transformations but require a good initialization. Finally, the matching
can be viewed as a discrete optimization problem with cost functions assigned
to node or edge pairings [10–12], which is very powerful but computationally
demanding. Pruning the search space is important to increase efficiency [4,5]
and node and edge descriptors [13,14] can be used for that purpose.
3 Problem Definition
Let us have an undirected graph GA = (VA , EA ) with nodes VA = {v1A ,
. . . , v|V
A
A | } and edges E
A
= {eA
1 , . . . , e|EA | }. We represent each edge ek ∈ E
A A A
4 Path Descriptors
The geometrical transformation between images in biomedical applications can
usually be decomposed into a rigid motion plus a mildly non-linear component.
Typically, the scale is known from the acquisition parameters and we can consider
it is equal to zero, without loss of generality. Due to mechanical properties of
the tissue, the nonlinear component is small and we therefore assume that for
all x, y ∈ RD , there is a bound εT on the relative length change, such that
1
d x, y ≤ d T (x), T (y) ≤ (1 + εT )d x, y , (1)
1 + εT
holds for a transformation T , where d(x, y) is the Euclidean distance and εT is
a deformation parameter, i.e. it is bi-Lipschitz.
We propose to use path descriptors – a vector characterizing each curve asso-
ciated with a geometric graph edge. Unlike previously suggested curve descrip-
tors [13,14,16], we have a good estimate of the change of descriptor values under
our transformation model. Let us have a sampling vector ω = (ω0 , . . . , ωnω +1 ),
such that 0 = ω0 < ω1 < · · · < ωnω < ωnω +1 = 1. Given a geometric path
ξsk : [0, 1] → Rd (Section 3) of a superedge sk , we calculate its descriptor param-
eterized by ω:
nω
nω
hω (sk ) = d ξsk (ωi ), ξsk (ωi+1 ) = ξsk (ωi+1 ) − ξsk (ωi ). (2)
i=0 i=0
compatible, i.e. if eq. (4) holds. Otherwise Wik;jl = 0. If matching node viA with
vkB is consistent with matching node vjA with vlB , Wik;jl is high and vice versa.
The matching is represented by a binary vector x∗ , such that xij = 1 iff
nodes viA and vjB match, maximizing the total affinity
A
||VB |
x∗ = arg max x Wx s.t. x ∈ [0, 1]|V , (5)
x
|U | |V |
∀j xij ≤ 1, ∀i xij ≤ 1.
i=1 j=1