presented to obtain the
Habilitation à Diriger des Reherhes
Université Paul Sabatier Toulouse 3
Mention: Applied Mathematis
by
Didier Auroux
Fast algorithms for image proessing and data
assimilation
Defended on 26 November 2008
After reviews by:
G.Aubert, Professor Université de NieSophia-Antipolis
P. Rouhon , Professor Mines ParisTeh
F. Santosa ,Professor University of Minnesota
Commitee members:
G.Aubert, Professor Université de NieSophia-Antipolis (Reviewer)
J. Blum , Professor Université de NieSophia-Antipolis (Examiner)
L. Cohen , Researh diretor CNRS & Université ParisDauphine (Examiner)
P. Degond , Researh diretor CNRS & Université PaulSabatier (Examiner)
M. Masmoudi , Professor Université PaulSabatier, Toulouse (Advisor)
J.-P. Puel , Professor Université de VersaillesSaint-Quentin (Examiner)
1 Introdution 7
2 Imageproessing 11
2.1 Introdution . . . 11
2.2 Topologial asymptotianalysis . . . 12
2.2.1 Presentation ofthemethod . . . 12
2.2.2 Mainresult . . . 12
2.3 Inpainting [9,13℄ . . . 13
2.3.1 Crakloalization problem . . . 14
2.3.2 Dirihlet and Neumannformulations forthe inpainting problem 14 2.3.3 Asymptoti expansion . . . 15
2.3.4 Algorithm . . . 16
2.3.5 Remarks . . . 17
2.4 Restoration [16, 22℄. . . 17
2.4.1 Variational formulation . . . 17
2.4.2 Topologial gradient . . . 18
2.4.3 Algorithm . . . 19
2.4.4 Remarks . . . 19
2.4.5 Extension to olorimages . . . 19
2.5 Classiation [10,16℄ . . . 21
2.5.1 Introdutionto the lassiation problem . . . 21
2.5.2 Restoration and lassiation oupling . . . 21
2.5.3 Extension to unsupervised lassiation . . . 22
2.6 Segmentation [12,13℄. . . 23
2.6.1 From restorationto segmentation . . . 23
2.6.2 Power seriesexpansion . . . 24
2.6.3 Algorithm . . . 25
2.7 Complexityand speedingup [13, 16℄ . . . 26
2.7.1 Disreteosine transform . . . 26
2.7.2 Preonditi onedonjugate gradient . . . 27
2.8 Topologial gradientand minimalpaths [26℄ . . . 27
2.8.1 Minimal paths . . . 28
2.8.2 Fast marhing. . . 28
2.8.3 Coupled algorithm . . . 29
2.9 Conlusions and perspetives . . . 30
3 Bak and forth nudging 31 3.1 Introdution . . . 31
3.2 BakandForthNudging (BFN)algorithm [8, 11,18℄ . . . 33
3.2.1 Forward nudging . . . 33
3.2.2 Bakward nudging . . . 34
3.2.3 BFN algorithm . . . 34
3.2.4 Choie ofthe nudgingmatriesand interpretation . . . 35
3.3 Numerial experiments [11,14,21℄ . . . 37
3.3.1 Numerial hoie of the nudgingmatries . . . 37
3.3.2 Experimentalapproah . . . 37
3.3.3 Physial models. . . 38
Lorenz equations . . . 38
Visous Burgers' equation . . . 38
Shallow watermodel . . . 39
Layered quasi-geostrophi oean model. . . 39
3.3.4 Conlusions emerging fromthe numerial experiments . . . 40
3.4 Theoretial onvergene results [18, 24℄ . . . 41
3.4.1 Linear ase . . . 41
3.4.2 Transportequations . . . 42
Linear visous transport . . . 42
Visous Burgers. . . 44
Invisid lineartransport . . . 45
Invisid Burgers . . . 46
3.5 Nudging and observers[25℄ . . . 47
3.5.1 Observers for ashallow watermodel . . . 48
3.5.2 Invariant orretion terms . . . 48
3.5.3 Convergene study ona linearized simpliedsystem . . . 49
3.5.4 Numerial experiments . . . 51
3.6 Conlusion. . . 52
4 Image data assimilation 53 4.1 Introdution . . . 53
4.2 Desription of the algorithm [23℄ . . . 54
4.2.1 Constant brightness assumption . . . 54
4.2.2 Cost funtion . . . 55
4.2.3 Regularization . . . 55
4.2.4 Muti-grid approah and optimizatio n . . . 55
4.2.5 Qualityestimate . . . 57
4.3 Numerial experiments [23℄ . . . 57
4.3.1 Simulated data . . . 57
4.4 Conlusions . . . 59
5 General onlusions and perspetives 61
List of publiations 63
Publiations linked to the PhD thesis. . . 63
Publiations afterthe PhD thesis . . . 64
Referenes 67
Introdution
In this work, several problems have been studied with the ommon goal of pro-
viding robust, partiularly easy to implement, fast and powerful algorithms. The
eieny of the algorithms is required by the operational ontext of the methods,
and by a need to proess more and more data in an inreasingly short time. The
other onstraint thatwe partiularly tookinto aount isthe easeof use and imple-
mentation of the methods we have developed.
In hapter 2, we will takle various problems in image proessing by an original
approah intheeld: topologialasymptotianalysis, ormoresimplythe topologial
gradient.
There has reently been a renewed interest in image proessing thanks to new
appliations in teleommuniations and mediine: on one hand, new tehnologie s
in teleommun iat ions and diusion of information, whih now involve sending and
reeiving massive ows of numerial data (e.g. images), and on the other hand the
medial world, in whih huge progress has been made, in partiular for the early
detetion oftumors, thanksto morepowerfulimaging tehniques.
Our study is motivated by several observations. First, the topologial gradient
is generallyusedfor strutural mehanis, design,and shape optimization problems.
Also,ithasbeen suessfullyapplied in eletromagnetism for the detetionofraks
or hiddenobjets. However,manyimage proessing problemsrelyon the good iden-
tiation of a subset of the image, for instane edges or harateristi objets. This
ommonfeatureseemedinterestingtous,andallowedustoadaptthetopologialgra-
dient method, initiallyusedfor rakdetetion, toseveralimage proessingproblems
(restoration, lassiation, segmentation, inpainting).
Theseondinteresting aspet isthe speedof the method. In various elds,topo-
logial asymptoti analysis hasmade it possible to obtain good results very quikly.
However,medial imagingand audiovisual diusion(e.g. satellite television orinter-
net broadasting) both requirethe proessingtime to be negligible. Ifthe proessing
time is too large, it will delay the medial diagnosis, or the ow of data. It is thus
important to buildextremely fastshemes for solving thesevarious problems,in real
time formoviesand anegligibletime (e.g. smaller than oneseond) for images.
Aswewillseehereafter,thetopologialgradientmethodatuallyadaptsperfetly
toimageproessingproblems,allowingustoobtainveryinterestingresultsforapar-
tiularlysmall omputation ost.
In hapter 3, we will study data assimilation for environmental and geophysial
problems, and more partiularly within the framework of atmospheri and oeani
observations. For several years, one of the major onerns has been to appreiably
improve our knowledge of these turbulent systems,one ofthe major goals being the
abilityto predit their evolution with ahighreliability.
Several dierent hallenges appear in data assimilation: short-range (e.g. a few
days)weatherforeasting,thestudyofglobalwarmingandlimatehange,detetion
of extreme limati phenomena several weeks in advane, ...For all these problems,
the goals are almost similar. They onsist of estimating quikly and with a very
high degree of auray the state of a turbulent system, from the ombined knowl-
edgeofmodels anddata: on onehandmathemati al equations modelingthe oupled
atmosphere-oeansystem,andontheotherhandobservationsofdierentnature(e.g.
in situ,or satelliteobservations), orresponding tovarious physial quantities.
Beyondthe extreme size ofthe problem tobesolved(several billionsofvalues to
be identied from hundreds of millions of observations)and the omputation al time
neededtosolve it,another fatorappears: the ostof development anduseofa data
assimilationmethod. Presently, itisextremelydiult toimplement suhamethod,
even on a relatively simple problem. This motivated us to study the possibility of
improving one ofthe simplestmethods of data assimilation, nudging (also known as
Newtonian relaxation), in order to obtain muh better results without ompliating
the method.
By applying the nudging method to the bakward (in time) problem, we noted
thatitis possible to stabilizethe bakward system, whihis unstablebeause of the
irreversibility of the physial problem. Thus, as detailed in hapter 3, we an go
bak in time, and obtain a more reliable estimate of the system at a previous time,
fromwhih foreasts maybe dedued. By applyingalternatively and repeatedly the
standard nudgingmethod to the forward and bakward models, we obtain an itera-
tive algorithm that is very easy to implement and provides denitely better results
than the standard nudging. Indeed, the results are of similar quality, and are often
obtainedmuhmorequiklythan byusingthestandardvariational dataassimilation
method.
Chapter4presentsastudyat theinterfae ofthesetwo elds: the assimilationof
images. Presently, ahuge quantityofobservations omingfromsatelliteimagesises-
sentiallynotusedto improve the knowledgeofthe systemstate. However,sequenes
ofimagesobtained bysatellitesdenitely showvarious harateristi strutures(hur-
rianes,swirls, urrentsof hot water, pollution, ...) moving and evolvingin time.
Several approahesan be onsideredto solvethis kindof problem,and wemade
Thatappearedtoustobethemostadaptedhoieforrapidlyextratingonventional
data (i.e. diretly related to the model variables), and then being able to use them
in a standard assimilationsystem.
Theideathatwedevelopinhapter4isbasedontheonstantbrightnessassump-
tion, whih onsists of looking for a displaement eld that transports an image to
another one. The originalityof our approah liesin the nonlinearization of the ost
funtiontobeminimized, ombinedwith afastmethodtoassembletheJaobianma-
trix. Finally, a multi-grid approah makes itpossible to guarantee the qualityofthe
minimum. Thanks to all these tehniques, we are able to extrat omplete veloity
elds in avery shorttime, anditis alsopossibleto provide aqualityestimateofthe
identied elds, whih an be viewed aserror statistisof these pseudo-observations
within the framework ofdata assimilation.
Finally some general onlusions and researh perspetives are given in hapter
5.
Image proessing by topologial
asymptoti analysis
Thishapter summarizes the work presented in [9, 10, 12, 13, 16, 22, 26℄.
2.1 Introdution
Theideaoftopologial asymptoti analysisisto measurethe impatofa pertur-
bationofthedomainonaostfuntion. Weonlyonsiderheretheapproahthathas
beenintrodued for topologialoptimizatio n purpose, in whihthe goalistoidentify
an optimalshape and itsomplement ar yin a givendomain [133, 98, 104℄.
Topologial shape optimizatio n seems partiularly well adapted to solve image
proessingproblems(likelassiation,segmentation, enhanement ,inpainting,...),
asthey mainly onsistof identifyinga partiular subdomain ofthe image: itsedges.
Atrstsight,themainissueoftopologialshapeanalysisisthenon-dierentiability
of the problem. To nd the optimal domain is indeed equivalent to identify its
harateristi funtion. Several lassial approahes have been developed to make
this problem dierentiable. We an ite here the relaxation tehnique, whih al-
lows the harateristi funtion to take all possible values in the interval
[0; 1]
, andthe level set approah where the harateristi funtion is replaed by a regular
level set funtion whih is positive inside the optimal domain and negative outside
[133, 34, 33, 36, 51, 157℄.
Theideaoftopologialasymptotianalysisistoswiththe harateristi funtion
fromonetozero(orfromzerotoone)ina(innitely)smallarea. Thus,thevariation
of the ost funtion is small when we swith a very small part from the subdomain
to its omplement ary. The topologial asymptoti expansionprovides this variation,
andallowsoneto deriveatopologial gradient oftheostfuntion [133,98, 158,157℄.
Inthis hapter, werstpresent thebasitoolsoftopologialasymptoti analysis,
and we then study several appliations to image proessing problems: inpainting
(where the goal is to ll a hidden part of an image), restoration and enhanement ,
lassiation, and segmentation. Then, we present a very eient way to speed up
allthealgorithmsintrodued inthishapter,basedondisreteosine transformsand
an appropriate preonditioning. Finally, we present a oupled approah ombining
thetopologialgradientandtheminimalpath tehniqueinordertoimprovethe edge
detetion, andto avoidnon-onnexontours.
2.2 Topologial asymptoti analysis
2.2.1 Presentation of the method
Let
Ω
bearegular openboundeddomain ofR 2 (orR 3). Letus onsideraPartial
Dierential Equation (PDE) problem dened in Ω
, written in its variational formu-
Ω
, written in its variational formu-lation:
nd
u ∈ V
suhthata(u, w) = l(w), ∀ w ∈ V ,
(2.1)where
V
isaHilbertspaeonΩ
,usuallyH 1 (Ω)
,a
isabilinearontinuousandoeriveformdened on
V
,andl
isalinearontinuousformonV
. Wenally onsideraostfuntion
J (Ω, u)
tobe minimized, whereu
isthe solution ofequation(2.1).We now onsider a small perturbation of the domain, e.g. by the insertion of a
rak
σ ρ = x 0 + ρσ(n)
,wherex 0 ∈ Ω
representsthepointwhere therakisinserted,σ(n)
is a straight rak ontaining the origin of the domain, andn
is a unit vetornormal to the rak. Finally,
ρ > 0
representsthe size of the perturbation ,assumed to be small. LetΩ ρ = Ω \ σ ρ be the perturbed domain. We an onsider the same
PDE problemasbefore, but on the perturbed domain:
nd
u ρ ∈ V ρsuhthata ρ (u ρ , w) = l ρ (w), ∀ w ∈ V ρ ,
(2.2)
where
V ρ, a ρ and l ρ represent the restrition of the Hilbert spae V
to Ω ρ, and the
l ρ represent the restrition of the Hilbert spae V
to Ω ρ, and the
perturbedbilinear and linearforms respetively.
We anrewrite theostfuntion
J
asafuntionofρ
byonsideringthe followingmap:
j : ρ 7→ Ω ρ 7→ u ρ solution of(2.2) 7→ j(ρ) := J(Ω ρ , u ρ ).
(2.3)
The topologial sensitivitytheory providesan asymptoti expansion of
j
whenρ
tendsto zero. Ittakesthe general form:
j(ρ) − j(0) = f(ρ)G(x 0 ) + o(f (ρ)),
(2.4)where
f (ρ)
is an expliit positive funtion going to zero withρ
, andG(x 0 )
is alledthe topologial gradient at point
x 0 [133℄.
Then to minimize the riterion
j
, one has to insert small holes (or raks) atpoints where the topologial gradient
G
is the most negative, in order to make theostfuntion
j
dereasequikly(see the asymptoti expansion(2.4)).2.2.2 Main result
Theorem 2.1 Ifthereexistsalinearform
L ρdenedonV ρ,afuntionf : R + → R +,
and fourreal numbers δJ 1, δJ 2,δa
andδl
suh that
f : R + → R +,
and fourreal numbers δJ 1, δJ 2,δa
andδl
suh that
δJ 2,δa
andδl
suh that
• lim
ρ→0 f (ρ) = 0,
• J (Ω ρ , u ρ ) − J (Ω ρ , u 0 ) = L ρ (u ρ − u 0 ) + f(ρ)δJ 1 + o(f (ρ)),
• J (Ω ρ , u 0 ) − J (Ω, u 0 ) = f (ρ)δJ 2 + o(f (ρ)),
• (a ρ − a 0 )(u 0 , p ρ ) = f (ρ)δa + o(f (ρ)),
• (l ρ − l 0 )(p ρ ) = f (ρ)δl + o(f (ρ))
,where the adjoint state
p ρ is the solutionof the adjoint equation
a ρ (w, p ρ ) = − L ρ (w), ∀ w ∈ V ρ ,
(2.5)and
u ρ is the solution of the diret equation (2.2), then the ost funtion j
has the
asymptoti expansion (2.4),where the topologial gradient
G(x)
isgiven byG(x) = δJ 1 + δJ 2 + δa − δl.
(2.6)2.3 Inpainting [9, 13℄
In this setion, we present an appliation of the topologial asymptoti analysis
to theinpaintingproblem. Thegoalofinpaintingistolla hiddenpartof animage.
In other words, if we denote by
Ω
the original image andω
the hidden part of theimage,thegoalistoreoverthehiddenpart
ω
fromtheknownpartoftheimageΩ \ ω
.There aremanyappliations, for instaneremoving somespotson abadly preserved
movie or image, ordeleting enrustedlogos and imageson television programs, ...
This problem has been widely studied. Several methods have been onsidered:
learningapproahes(neuralnetworks,radialbasisfuntions,supportvetormahine,
...),in whih the learning data istaken in
Ω \ ω
, and then the approximate funtion is evaluated inω
[177, 178℄; minimization of an energy ost funtion inω
based ona total variation norm[67, 68℄; morphologial analysisfor the reonstrutionof both
artoonand texture[87℄; ...
In order to study the inpainting problem, we rst onsider a rak loalizatio n
method. Crak detetion allows us to identify the edges of the hidden part of the
image, and the inpainting problem an then be easily solved. We will onsider the
lassial thermal diusiontehnique [142, 66, 174, 175, 150℄ and improve itbymod-
eling the edges byraks. Theseraks aresupposed to be highly insulating and to
allow the temperature to jump aross edges. As both the Dirihlet and Neumann
onditionsareknownonthe boundaryofthe hiddensubset,wean deneariterion
measuringthe disrepanybetween thesolutionsofaDirihletandaNeumannprob-
lem respetively [118℄. This problem is similar to the inverse ondutivity problem,
ofapartialdierentialequationfromtheknowledgeoftheDirihlettoNeumannoper-
ator. Onlytwomeasurementsareneededtoreoverseveralsimpleraks[30,31,48℄.
Fromthe numerialpoint ofview,severalmethods [40, 49, 50, 64, 96, 152, 151℄ have
been proposed, but the topologial gradient approah seemsto be the most eient
methodfor rakloalizatio n. Theminimization ofthe riterionallowsus to identify
themainedgesinsidethehiddenpartoftheimage. Theimageisnallylledbetween
the edgesthanksto the Laplae operator.
This setion summarizes the work introdued in [9, 13℄. We also refer to these
referenesfor the results ofmanynumerial experiments.
2.3.1 Crak loalization problem
Let
Ω
be abounded open setofR 2. We assumein this setionthatΩ
ontains a
perfetly insulating rak
σ ∗. We impose a uxφ ∈ H −1/2 (Γ)
on the boundary Γ
of
Ω
, and we want to ndσ ⊂ Ω
suh thatthe solution u ∈ H 1 (Ω \ σ)
of
∆u = 0 in Ω \ σ,
∂ n u = φ on Γ,
∂ n u = 0 on σ,
(2.7)
satises
u | Γ = T
, whereT ∈ H 1/2 (Γ)
is a given funtion. We also assume someompatibilityonditions in orderto have awell-poseddiret problem.
Atopologialgradientapproahhasbeenintrodued in[37℄,andonsistsofden-
ing a Dirihlet and a Neumann problem, as we have an over-determina tion in the
boundary onditions:
u D ∈ H 1 (Ω \ σ)
suhthat
∆u D = 0 in Ω \ σ, u D = T on Γ,
∂ n u D = 0 on σ,
(2.8)
u N ∈ H 1 (Ω \ σ)
suhthat
∆u N = 0 in Ω \ σ,
∂ n u N = φ on Γ,
∂ n u N = 0 on σ.
(2.9)
Itislear thatfortheatualrak
σ ∗,the twosolutionu D andu N areequal. The
u N areequal. The
ideaisthen to onsider andminimize the following ostfuntion
J (σ) = 1
2 k u D − u N k 2 L 2 (Ω) .
(2.10)Thetopologial asymptoti expansionof this ostfuntion isdetailed in [37℄.
2.3.2 Dirihlet and Neumann formulations for the inpainting prob-
lem
In our approah, we now denote by
Ω
the image andΓ
its boundary,ω ⊂ Ω
themissingpart of the image and
γ
its boundary. Letv
be the image that we want torestore. Weassume that
v
isknown inΩ \ ω
, and unknown inω
.Theideaisto adapt the rakloalization method to inpainting: rakdetetion
rstallowsustoidentifytheraks(oredges)
σ
ofthehiddenpartω
oftheimage,andthen we willimposethattheLaplaianof therestored imageisequal to zeroin
ω \ σ
.For agivenrak
σ ⊂ ω
,asv
(Dirihletondition) and∂ n v
(Neumann ondition)areknown onthe boundary
γ
ofω
, we ansolvetwo dierent problems insideω
.For a given rak
σ
, we denote byu D ∈ H 1 (Ω \ σ)
the solution of the followingDirihletproblem:
∆u D = 0 in ω \ σ, u D = v on γ,
∂ n u D = 0 on σ, u D = v in Ω \ ω.
(2.11)
Outside
ω
, the solution isequal to the original image, andinsideω
, we useequation(2.8).
Inthesameway,ifweassume
v
tobeenoughregular,weanonsiderthesolutionu N ∈ H 1 (Ω \ σ)
ofthe following Neumann problem:
∆u N = 0 in ω \ σ,
∂ n u N = ∂ n v on γ,
∂ n u N = 0 on σ, u N = v in Ω \ ω.
(2.12)
Notethatfromthe numerial point ofview,itismuhmoreeasyto solvean approx-
imatedNeumann problem:
∆u N = 0 in ω \ σ,
∂ n u N = ∂ n v on γ,
∂ n u N = 0 on σ,
− α∆u N + u N = v in Ω \ ω,
(2.13)
where
α
isa small positive number.2.3.3 Asymptoti expansion
The ost funtion remains unhanged, and is still dened by (2.10), as the idea
is to nd some raks
σ ⊂ ω
thatminimize the dierene between the two solutionsu N and u D. We assume thatthe rakσ
is equal to x + ρσ
, where x
is the point of
σ
is equal tox + ρσ
, wherex
is the point ofinsertion oftherak,
ρ
isthe sizeof theinsertedrak(assumedto be small),andσ
is areferene rak, of unitnormal vetor
n
. Then, we anrewrite the ost funtionJ
dened by equation (2.10) as a funtionj(ρ)
ofρ
. The asymptoti expansion isthen the following:
j(ρ) − j(0) = f (ρ)g(x, n) + o(f (ρ)),
(2.14)where the topologial gradient
g
is dened byg(x, n) = − [( ∇ u D (x).n)( ∇ p D (x).n) + ( ∇ u N (x).n)( ∇ p N (x).n)] ,
(2.15)where
u D and u N are the solutions of (2.11) and (2.12) respetively, but without
any inserted rak (σ = ∅
). Also, p D and p N are the orresponding adjoint states,
respetively solutions in H 1 (Ω)
of the following equations:
σ = ∅
). Also,p D and p N are the orresponding adjoint states,
respetively solutions in H 1 (Ω)
of the following equations:
H 1 (Ω)
of the following equations:
p D = 0 in Ω \ ω, p D = 0 on γ,
− ∆p D = − (u D − u N ) in ω,
(2.16)
p N = 0 in Ω \ ω,
∂ n p N = 0 on γ,
− ∆p N = +(u D − u N ) in ω.
(2.17)
The topologial gradient dened by equation (2.15) an be rewritten in the fol-
lowing way:
g(x, n) = n T M (x)n,
(2.18)where
M (x)
is the2 × 2
(resp.3 × 3
in the aseof3Dimages, or movies) symmetrimatrixdened by
M(x) = − sym( ∇ u D (x) ⊗ ∇ p D (x) + ∇ u N (x) ⊗ ∇ p N (x)).
(2.19)From this equation, we an dedue thatthe minimum of
g(x, n)
is reahed whenn
isthe eigenvetor assoiated to the lowesteigenvalueλ min (M(x))
ofM(x)
.2.3.4 Algorithm
Theinpainting algorithm isthen the following:
•
Calulation ofu D and u N, solutions of the diret problems (2.11) and (2.12)
respetively, without anyinsertedrak(unperturbed problem:
σ = ∅
).•
Calulation ofp D and p N the two orresponding adjoint states, respetively
solutions of equations (2.16) and (2.17).
•
Computation ofthe matrixM (x)
dened byequation (2.19).•
Loalizati onof the raks: deneσ = { x ∈ ω; λ min (M (x)) < δ < 0 } ,
(2.20)where
δ
isa negative threshold.•
Calulation of the solution of the Neumann problem (2.12) perturbed by theinsertion of
σ
.Thisimage isthen equal to theoriginal image in
Ω \ ω
, andithasbeenreonstruted inω
.2.3.5 Remarks
From the numerial point of view, raks are modeled by a very small ondu-
tivityinstead of onsidering real holes in the domain. The previous algorithm has a
omplexityof
O (n. log(n))
,wheren
isthesizeoftheimage,i.e. thenumberofpixels,asexplained in setion2.7.
The main advantage of this algorithm is that the reonstrution is done in only
one iteration of the topologial gradient algorithm, whih onsistsof
5
resolutions of a PDE (the two diret and two adjoint unperturbed problems, and then one diretperturbedproblem)inthedomain
Ω
representingtheimage. Severalnumerialresults arepresented in [9℄ andshow the qualityand eieny of the reonstrution.Theonlyontrolparameterofthismethodisthenegativethreshold: belowagiven
value,thepixelsareonsideredasbeingpartoftheedgeset,whereasitisnotthease
beyond the threshold. The reonstruted image isprovided by the resolution of the
diret perturbed problem(2.12),andthe qualityof theimage relies onthe onnexity
of the identied edges. Ifa givenidentiededge isnot onnex,the Laplaian indeed
produes a blurred zone. Then, the threshold is usually set suh that the main
identied edges are onnex. Of ourse, it may lead to the wrong identiation of
edges. But the various numerial experiments have shown that the threshold an
be xed to an a priori value, as the optimal threshold is almost independent of the
images.
Anothersolution to thisproblem ispresented in setion2.8.
2.4 Restoration [16, 22℄
In this setion, we onsider the restoration problem, with the aim of restoring
noisyimages. Themainideaisto usethetopologialgradientfor detetingthe edges
of the noisyimage in orderto preserve themduring the restorationproess.
Thismethod is based onthermal diusion, like manyother variational methods.
Inordertoavoidblurringeets, severalnonlinearisotropi andanisotropimethods
havebeenintrodued,someofthemrelyingontheminimizationofthetotalvariation
[142,66,124, 174,175,43℄. Weshouldmentionthatsomenonvariational approahes
also exist,mainly statistialmethods [86℄.
This setion summarizes the work presented in [16, 22℄. We also refer to these
referenes for the resultsof numerial experiments.
2.4.1 Variational formulati on
Let
Ω ⊂ R 2 bean openboundeddomain,and v ∈ L 2 (Ω)
bethe noisyimage. The
enhanement of
v
isbasedon the resolution of the following problem:nd
u ∈ H 1 (Ω)
suhthat− div(c ∇ u) + u = v in Ω,
∂ n u = 0 on ∂Ω,
(2.21)where
n
is the outward unit normal to∂Ω
, andc
is the ondutivity, to be de- ned in the following. Several hoies an be made for the ondutivity, mainlyc
equal to a onstant value (linear diusion method: it is fast, but it blurs important
strutures), or
c
dened bya nonlinear funtion of∇ u
(nonlinear diusion method,edge-preserving [175, 43℄). In the topologial gradient approah,
c
takes only twovalues: a onstant value
c 0 (loseto 1
) in the smooth part of the image, and a very
smallvalues
ε
(loseto0
) onthe edges or raks in orderto preservethem.Setting
c = 0
on a part of the image is equivalent to perturbing the domain bythe insertion of raks. For a given point
x 0 ∈ Ω
and for a given small parameterρ > 0
, we onsiderΩ ρ = Ω \ σ ρ the perturbed domain by the insertion of a rak
σ ρ = x 0 + ρσ(n)
, where σ(n)
isa straight rakand n
is aunit vetor normal to the
rak. The variational formulation of the perturbed problemis the following:
nd
u ρ ∈ H 1 (Ω ρ )
suh thata ρ (u ρ , w) = l ρ (w), ∀ w ∈ H 1 (Ω ρ ),
(2.22)where
a ρ (resp. l ρ) is the following bilinear (resp. linear) form dened on H 1 (Ω ρ )
H 1 (Ω ρ )
(resp.
L 2 (Ω ρ )
)bya ρ (u, w) =
Z
Ω ρ
(c ∇ u ∇ w + uw) dx, l ρ (w) = Z
Ω ρ
vw dx.
(2.23)Edge detetion ifequivalent to looking for asubdomain of
Ω
where the energy issmall. So our goalis to minimizethe energy norm outsideedges:
j(ρ) = J (Ω ρ , u ρ ) = Z
Ω ρ
k∇ u ρ k 2 .
(2.24)2.4.2 Topologial gradient
From theorem 2.1, we an derive the following asymptoti expansion of the ost
funtion(2.24):
j(ρ) − j(0) = ρ 2 G(x 0 , n) + o(ρ 2 ),
(2.25)where
G(x 0 , n) = − πc( ∇ u 0 (x 0 ).n)( ∇ p 0 (x 0 ).n) − π |∇ u 0 (x 0 ).n | 2 ,
(2.26)where
p 0 is the solutionof the unperturbed adjoint problem:
− div(c ∇ p 0 ) + p 0 = − ∂ u J (Ω, u 0 ) in Ω,
∂ n p 0 = 0 on ∂Ω.
(2.27)Aspreviouslyseen,thetopologialgradientanberewritten:
G(x, n) = h M (x)n, n i
,where
M(x)
isthe following2 × 2
symmetrimatrix:M(x) = − πc ∇ u 0 (x) ∇ p 0 (x) T + ∇ p 0 (x) ∇ u 0 (x) T
2 − π ∇ u 0 (x) ∇ u 0 (x) T .
(2.28)2.4.3 Algorithm
Our algorithm onsists of inserting small heterogenei ti es (or raks) in regions
where the topologial gradient is smaller than a given threshold. There regions are
the edges ofthe image. Thealgorithm isasfollows:
•
Initialization:c = c 0 (onstant valueeverywhere).
•
Calulation ofu 0 and p 0, respetivelysolutions of the diret (2.21) andadjoint
(2.27) unperturbed problems.
•
Computation of the2 × 2
matrixM (x)
dened by (2.28), and of its lowesteigenvalue
λ min (M(x))
at eah point of the domain.•
Set the new ondutivity:c 1 =
ε
ifx ∈ Ω
is suhthatλ min (M(x)) < α < 0,
c 0 elsewhere, (2.29)
where
ε > 0
isassumed to be small,andα
is anegative threshold.•
Calulation ofu 1, the solution of the perturbed diret problem (2.21) using
c = c 1.
Theimage
u 1 isthe restored image.
2.4.4 Remarks
Fromthenumerialpointofview,itismoreonvenienttosimulatetheraksbya
smallvalueof
c
insteadof onsideringtopologial perturbation sofΩ
. Theresolutionof problem(2.21) with
c = c 1 is anapproximation ofthe resolution ofthe perturbed
problem(2.22),beoming morepreise asε
goesto 0
.
Asin the previous setion (inpainting problems), our algorithm is extremely ef-
ient as it requires only
3
resolutions of a partial dierential equation inΩ
: thediret and adjoint original problems, and then the diret perturbed problem. And
the omplexityofthis algorithm is still
O (n. log(n))
(seesetion 2.7).Asshown in [16℄, the quality of the numerial results is very good. One again,
thealgorithmreliesonathresholdingofthetopologialgradientinordertodenethe
edge set. Contrary to inpainting problems, the onnexity of the edges isnot ruial
sine it does not hange signiantly the quality of the restored image. However,
setion2.8presentsawaytoidentifyonnexedges, withfewer badlyidentiededges.
2.4.5 Extension to olor images
Inthissetion, weadaptthe topologialgradient approahtoolorimages. Color
images an be represented or modeled in various ways, for instane the RGB (Red-
Green-Blue) spae in whih images are viewed asfuntions from
Ω
toR 3 instead of
R
. A rst approah onsists of deoupling the three hannels, and in solving diretandadjoint problemsforeahhannel. Butitisalsopossible toonsiderdiretlythe
vetorial minimization problem, involving the resolution of vetorial problems. The
topologial asymptoti expansion is still given by equations (2.25-2.26) and (2.28),
whereallfuntionsarevetorial,i.e. thetopologialgradientisthesumonallhannels
ofthe orresponding expressionsfor eah hannel[22℄.
Another approah hasalsobeenstudied in [22℄,in whih we useadierent norm
for oupling the dierent hannels. In order to identify the loal variations of the
olor image, Di Zenzo denes a multi-spetraltensor assoiated to the image vetor
eld[83℄:
T =
t 11 t 12 t 21 t 22
, t ij = X 3
k=1
∂u k
∂x i
∂u k
∂x j , 1 ≤ i, j ≤ 2,
(2.30)in the ase of bidimensional images. This tensor desribes the rst orderdierential
struture ofthe image, andthe DiZenzo gradient isgiven bythe square root of the
largesteigenvalue ofthe struture tensor:
k∇ u k DZ = 1
√ 2
t 11 + t 22 + q
(t 11 − t 22 ) 2 + 4t 2 12 1 2
.
(2.31)Itis possible to rewrite this gradient in a dierent waywith the following funtion:
H( ∇ u) = 1 + p
1 − 4f ( ∇ u)
2 ,
(2.32)where
f ( ∇ u) = det 2 ( ∇ u 1 , ∇ u 2 ) + det 2 ( ∇ u 1 , ∇ u 3 ) + det 2 ( ∇ u 2 , ∇ u 3 )
( |∇ u 1 | 2 + |∇ u 2 | 2 + |∇ u 3 | 2 ) 2 ,
(2.33)det 2 ( ∇ u s , ∇ u t ) = ∂u s
∂x 1
∂u t
∂x 2 − ∂u t
∂x 1
∂u s
∂x 2 2
.
(2.34)Then,weanderivetheasymptotiexpansionoftheostfuntiondenedbyequation
(2.24) in whihthe norm isthe DiZenzo norm (2.31):
G(x 0 , n) = X 3
k=1
h − πc( ∇ u k 0 (x 0 ).n)( ∇ v k 0 (x 0 ).n) − πH( ∇ u 0 (x 0 )) |∇ u k 0 (x 0 ).n | 2 i
(2.35)
with our standardnotations.
In [22℄, we showthat this approah has the same omputation al ost asthe ve-
torialapproah(inwhihthe dierent hannels aredeoupled), whileitimprovesthe
edge detetion, and hene it produes a better restored image, more preise on the
2.5 Classiation [10, 16℄
Inthis setion, we now fouson the regularized and unsupervised image lassi-
ation problem.
Inspiredbytheworkpresented in [150, 43℄,in whihtheauthors proposealassi-
ation modeloupledwith a restorationproess,we adapthereour approahbased
on the topologialasymptoti analysis.
Thissetionsummarizestheworkpresentedin[16,10℄. Werefertothesereferenes
for the numerial results.
2.5.1 Introdution to the lassiati on problem
Let
v
betheoriginalimagedenedonanopensetΩ
ofR 2,andletC i , 1 ≤ i ≤ n
,be
n
lasses(i.e. greyorolorlevels). Werstassumethatthesselassesarepredened.
Thegoal ofimage lassiation isto nda partitionof
Ω
in subsets{ Ω i } i=1...n, suh
that
v
is lose toC i in Ω i.
A variational approah an be dened: it onsists of a ost funtion measuring
the dierene between the original image andthe lassied image:
J((Ω i ) i=1...n ) = X n
i=1
Z
Ω i
(v(x) − C i ) 2 dx + α X
i6=j
| Γ ij | ,
(2.36)where
Γ ij representsthe interfae Ω i ∩ Ω j between two subsets.
Themain diultyof this approah isthat the unknowns aresets, and not vari-
ables. This is why the topologial asymptoti analysis seems to be appropriate for
solving this problem. The topologial gradient and the orresponding numerial re-
sultsarepresented in [10℄.
2.5.2 Restoration and lassiation oupling
Anothersolutiononsistsofouplinglassiation with restoration,and toadapt
theapproahintrodued insetion2.4. Theideaistorstonsideraniterationofthe
topologialasymptoti analysisfor theimage restorationprobleminorderto smooth
the image, and then to lassify this smooth image without anyregularizati on. Ifwe
removetheregularizati ontermfromequation(2.36),whihleadstotheunregularized
lassiation problem, then the optimal subset
Ω i is the set of pixels that areloser
to
C i than to any other C j. In other words, eah pixel is assigned to the subset
orresponding to itslosest lass.
In the perturbed problem (2.29), instead of setting
c = 0
(orc = ε
from thenumerial point of view) onthe edge set and
c = c 0 elsewhere, we set
c 1 =
( ε
on the edge set,c 0
ε
elsewhere.(2.37)
•
Appliation of the restorationalgorithm dened in setion 2.4, withc 1 dened
by(2.37) insteadof (2.29).
•
Unregularized lassiation ofthe imageu 1, usingfor examplethe losest lass
algorithm (in whih eah pixel is assigned to the subset orresponding to its
losest lass).
Aspreviouslyseen,the omplexityofthisalgorithm is
O (n. log(n))
, andthevari-ousnumerialresultspresentedin[10℄showtherelativeeienyoftheseapproahes.
Moreover,itispossible toregularize moreorlessthe imagebyhoosingdierent val-
uesof
c 1, and itallows us toalso obtain good resultson noisyimages.
2.5.3 Extension to unsupervised lassiati on
If the number
n
of lasses is given, but not their valuesC i, it is possible to
determinethem in an optimalway. Thislassiation probleman be dened as
(Ω min i ),(C i ) j((Ω i ), (C i )) = X n
i=1
Z
Ω i
| v(x) − C i | 2 dx + α X
i6=j
| Γ ij | .
(2.38)Theideaistominimizetheostfuntion
j((Ω i ), (C i ))
alternativelywithrespettoΩ i andwithrespettoC i. Theminimizationwith respettoΩ i onsistsoflassifying
Ω i onsistsoflassifying
the image, while the minimization with respet to
C i isobtained straightforward by the meanof the image in eahlass:
C i = 1
| Ω i | Z
Ω i
v(x) dx.
(2.39)Theunsupervised lassiation algorithm is then asfollows:
•
Initialization: dene an initial guessC 1 , . . . , C n (e.g. equi-distributed lasses).
•
Repeatuntil onvergene:Calulate the lassied image using the lasses
C 1 , . . . , C n (see previous
algorithm).
Updatethe values ofthe lasses using (2.39).
If the number
n
of lasses is not given, we an add a penalizatio n term+βn
in the ostfuntion (2.38), measuringthe number oflasses. Theminimization with
respetto
n
providesthe optimalnumber oflasses. The number oflasses islearlyrelatedto the hoie of the weighting oeient
β
.2.6 Segmentation [12, 13℄
Thissetionisonernedwithimagesegmentation,whihaimistondapartition
ofanimageintoitsonstituentparts. Theideaisstilltoapplyourtopologialgradient
basedalgorithm forthe detetion ofedges in the image.
Several approahes have been studied in the literature. One an ite variational
methods, for example based on the minimization of the Mumford-Shah funtional
[136℄,theativeontoursandsnakemethods [55,156℄,stohastiapproahes[54, 61℄,
wavelets, ...[43, 45, 42, 140, 149, 150, 176℄.
This setion summarizes the main results presented in [12, 13℄. Several numer-
ial experiments are also detailed in these referenes and show the eieny of our
approah.
2.6.1 From restoration to segmentation
Westill onsiderthe restorationalgorithm,in whihthe following ondutivityis
usedfor the perturbed problem:
c(ε) =
( ε
inω, 1
ε
outsideω,
(2.40)where
ω ⊂ Ω
represents the edge set. We rst assume thatω
is thikened (i.e. ofodimension
0
inΩ
). Fromequation(2.40),the algorithmnowonsistsofsolvingthefollowing problem:
( P ε )
− div(ε ∇ u ε ) + u ε = v in ω,
− div 1
ε ∇ u ε
+ u ε = v in Ω \ ω,
∂ n u ε = 0 on ∂Ω,
(2.41)
where
u ε ∈ H 1 (Ω)
, i.e. with the impliit boundary ondition thatc(ε)∂ n u ε hasthe
same valueon both sidesof
∂ω
.Thenwe havethe following asymptoti result[12℄:
Theorem 2.2 If wedenoteby
u εthe uniquesolutionofproblem ( P ε )
inH 1 (Ω)
, then
ε→0 lim ( k∇ u ε − ∇ u 0 k L 2 (Ω\ω) + k u ε − u 0 k L 2 (ω) ) = 0, (2.42)
where
u 0 ∈ H 1 (Ω \ ω) ∩ L 2 (Ω)
isthe solution tothe following problem( P 0 )
u 0 = v in ω,
− div ( ∇ u 0 ) = 0 in Ω \ ω,
∂ n u 0 = 0 on ∂ω,
∂ n u 0 = 0 on ∂Ω.
(2.43)
This result proves that the segmented image
u 0 an be approximated by u ε if ε
ε
is small. We now assume that the edge set
ω
is of odimension1
inΩ
. From thepoint of view of appliations, it is ompletely natural to assume that the edges are
atintheimage. Inordertohaveoherentnotations, wewillfurtherdenoteby
σ
theedgeset. Weassumethat
σ
isknown, e.g. provided bytherakdetetionalgorithmpreviouslyseen.
We an rewrite the approximated segmentation problem
( P ε )
asfollows:( ˜ P ε )
− div 1
ε ∇ u ε
+ u ε = v in Ω \ σ,
∂ n u ε = 0 on σ,
∂ n u ε = 0 on ∂ Ω,
(2.44)
where
u ε ∈ H 1 (Ω \ σ)
. Ifv ∈ L 2 (Ω)
, then problem( ˜ P ε )
has a unique solution inH 1 (Ω \ σ)
. Asa orollaryof the previousresult, we have the following one [12℄:Theorem 2.3 If we denote by
u ε the unique solution of problem ( ˜ P ε )
in H 1 (Ω \ σ)
,
then
k u ε k L 2 (Ω) ≤ k v k L 2 (Ω) , k∇ u ε k L 2 (Ω\σ) ≤ √
ε k v k L 2 (Ω) ,
(2.45)and
ε→0 lim k∇ u ε − ∇ u 0 k L 2 (Ω\σ) = 0, (2.46)
where
u 0 ∈ H 1 (Ω \ σ)
is the unique solutionto the following problem:( ˜ P 0 )
− div ( ∇ u 0 ) = 0 in Ω \ σ, Z
Ω i
u 0 = Z
Ω i
v ∀ Ω i onnex omponent of Ω \ σ,
∂ n u 0 = 0 on σ,
∂ n u 0 = 0 on ∂ Ω.
(2.47)
For numerial reasons, itan be verydiult tosolvediretlyproblem
( ˜ P 0 )
, andevenproblem
( ˜ P ε )
fortoosmallvaluesofε > 0
. Indeedtheonditioningofthesystem to be solved goes to innity whenε → 0
. In order to overome this issue, we willexpandthe solution
u ε of problem( ˜ P ε )
into a power series ofε
.
2.6.2 Power series expansion
From the knowledge of the power series expansion of
u ε and the omputation of
several solutions
u ε for not toosmall oeientsε > 0
, itispossible to approximate
the asymptotisolution u 0 [12℄:
Theorem 2.4 There exist a onstant
ε R > 0
and a family of funtions(u n ) n∈ N of
H 1 (Ω \ σ)
suh that for all0 ≤ ε ≤ ε R,
u ε = X ∞
n=0
u n ε n .
(2.48)Moreover,
u 0 isthe unique solutionin H 1 (Ω \ σ)
of problem ( ˜ P 0 )
,andthe other fun-
tions
(u n )
are the unique solutions inH 1 (Ω \ σ)
of the following problems:( ˜ P 1 )
− div ( ∇ u 1 ) = − u 0 + v in Ω \ σ, Z
Ω i
u 1 = 0 ∀ Ω i onnex omponent of Ω \ σ,
∂ n u 1 = 0 on σ,
∂ n u 1 = 0 on ∂Ω,
(2.49)
n ≥ 2, ( ˜ P n )
− div ( ∇ u n ) = − u n−1 in Ω \ σ, Z
Ω i
u n = 0 ∀ Ω i onnex omponent of Ω \ σ,
∂ n u n = 0 on σ,
∂ n u n = 0 on ∂Ω.
(2.50)
We andene a funtion of
ε ∈ R + asfollows
f (ε) := u ε ∈ H 1 (Ω \ σ).
(2.51)From the previous theorem, we know that
f
has a power series expansion at theorigin given by (2.48). We onsider a family of
N
points(ε i )
in[ε c , ε R ]
, whereε c is
the smallest value of
ε
for whih it is easy to numerially omputef (ε)
, andε R is
smaller than the onvergene radius of the power series. We an then ompute an
interpolatio n polynomial
g N of degreeN − 1
dened by:
g N (ε) = X N
i=1
Y N
j=1,j6=i
ε − ε j ε i − ε j
u ε i ,
(2.52)where
N
isthe number of pointsε i.
Theanalyityof
f
allowsus to estimatethe approximation error:k u 0 − g N (0) k H 1 (Ω\σ) = O (ε N c ).
(2.53)2.6.3 Algorithm
Wean thendene asegmentation algorithm,basedon therestorationalgorithm
previously dened in setion2.4:
•
Solve the diret (2.21) and adjoint (2.27) unperturbed problems withc = c 0
everywhere.
•
Compute the2 × 2
matrixM (x)
dened by equation (2.28) and its lowesteigenvalue
λ min (M(x))
at eah point of the domainΩ
.•
Deneσ = { x ∈ Ω; λ min < α < 0 }
the edge set, whereα
is a small negative•
Setε c > 0
the minimal value ofε
for whih it is easy to ompute numeriallythe solution
u ε of problem( ˜ P ε )
.
•
ChooseN ∈ N ∗ in order to have an approximationerror in O (ε N c )
, and hoose
N
dierent values(ε i )
.
•
Compute the solutions(u ε i )
inH 1 (Ω \ σ)
of problems( ˜ P ε i )
.•
Compute theinterpolatio n polynomialg N ofdegreeN − 1
,dened byequation
(2.52),for
ε = 0
.Thisalgorithmhasaomplexityin
O (N.n. log(n))
,wheren
isthenumberofpixelsin the image, and
N
is the degree of the interpolatio n approximation. In numerial experiments,N
is typiallyof the orderof2
to5
.Several numerial testsaredetailedin [12℄.
2.7 Complexity and speeding up [13, 16℄
Inthis setion, we present the tehniques thatwe have usedfor solving the PDE
problems previously seen, and that lead to a theoretial omplexity in
O (n. log(n))
[13℄. Several numerial experimentshave onrmed thisomplexity[16, 13℄.
2.7.1 Disrete osine transform
Inall the algorithmswepresented in the previous setions,we only have to solve
the following PDE
− div(c ∇ u) + u = v in Ω,
∂ n u = 0 on ∂Ω,
(2.54)forvarious oeients
c
. Therstresolutions aredonewith a onstant valueofc
. Itisthen possibletolargely speeduptheomputationtime byusingthe disreteosine
transform(DCT) method. Problem(2.54) isthen equivalent to
X
m,n
1 + c(mπ) 2 + c(nπ) 2
u m,n φ m,n = X
m,n
v m,n φ m,n ,
(2.55)where we denote by
φ m,n = δ m,n cos(mπx) cos(nπy)
a osine basisofR 2, and where
(v m,n )
representtheDCToeientsoftheoriginalimagev
. Itisthenstraightforward toidentify(u m,n )
, the DCToeients ofu
in equation(2.55):u m,n = v m,n
1 + c(mπ) 2 + c(nπ) 2 .
(2.56)Theomplexityofsuharesolutionis
O (n. log(n))
,wheren
isthenumberofpixelsofthe image. The resolution of all unperturbed problems is then done in the following
way:
•
Computation ofv m,n, the DCToeientsof the original image v
.
•
Computationofu m,n, the DCToeientsof u
fromequation(2.56).
•
Computationofu
usingan inverse DCT.2.7.2 Preonditioned onjugate gradient
Then,thesolutionofallpreviouslydetailedproblemsomesfromtheresolutionof
aperturbed problem. Forthe lastresolution ofa diretproblemwith anon onstant
oeient
c
, we an rewrite the problemin the following way:A(c)u = B,
(2.57)where
u
is the unknown image. Ifc
is onstant, equation (2.57) is easy to solve.The idea is to preondition equation (2.57) with the DCT solver used in the rst
resolution. Problem(2.57) isequivalent to
A(c 0 ) −1 A(c) u =
A(c 0 ) −1 B
.
(2.58)As
c
islose toc 0 (c
isindeedequal to c 0,exeptin anegligible partofthe domain),
the systemmatrix
[A(c 0 ) −1 A(c)]
is lose to the identityoperator,and the resolutionof (2.58) isthen easy: weusea preonditioned onjugate gradient (PCG)methodto
solve this problem. As the oeient
c
is lose toc 0, we an expet a O (n. log(n))
omplexityfor the resolution of the perturbed problem. The numerial experiments
learly onrm this omplexity, both for smalland large problems.
The main advantage is that it allows us to proess images in a very short time
(e.g.
1600 × 1200
images in lessthan one seond) and movies in real time (providedthe movie is splitinto shortsequenes of afew seonds)with a ++ode.
2.8 Coupling between the topologial gradient and the
minimal path tehnique [26℄
As previously seen, e.g. in setion 2.3, it is ruial to identify onneted (or
ontinuous) ontours. Up to now, we had to threshold the topologial gradient with
a nottoo smallvalue,in order toidentifyonneted ontours,but thisleads to thik
identied edges, and alsoto onsider morenoisypointsaspotential edges.
We notied thatthe edges orrespond to valleylines of the topologial gradient.
It isofourse possibleto identify them byadapting the thresholdoeient, butwe
propose here to use the minimal path and fast marhing tehniques for identifying
the valleylinesof the topologial gradient [71, 73, 82, 84, 179, 145, 164℄.
Inthe following,we onsideranyofthe previousimage proessing problems. We
only assume that the topologial gradient
g
has been dened and omputed every-where. The goal is to identify the valley lines orresponding to the most negative
parts ofthe topologial gradient.
Thissetion summarizes the study presented in [26℄, in whih several numerial