Biomedical Engineering Reference
In-Depth Information
Fig. 13.2. Definitions for a conformal mapping u
(
x
)
of a 3D surface
. The boundary of
is a
S
S
closed curve and is noted
S 0
as possible in the parametrization and have the same norm. This can bee seen as an
approximation of the Cauchy-Riemann equations. For a piecewise linear mapping,
the least square conformal map can be obtained by minimizing the conformal energy:
v
1
2
2
u
E LSCM (
u
)=
ds
,
(13.2)
S
where denotes a counterclockwise 90 rotation in
. For a 3D surface with normal
vector n , the counterclockwise rotation of the gradient can be written as:
S
u =
n
×
u (see Fig. 13.2). Eq. (13.2) can be simplified and rewritten as follows:
v ds
1
2
u ·
u +
u ·
E LSCM (
u
)=
v
·
v
2
,
S
,
(13.3)
1
2 (
=
u
·
u
+
v
·
v
2
(
n
×
u
) ·
v
)
ds
.
S
Recalling that a “dot” and a “cross” can be interchanged without changing the result,
we have:
1
2 (
E LSCM (
u
)=
u
·
u
+
v
·
v
2 n
· (
u
×
v
))
ds
.
(13.4)
S
We now derive the finite element formulation of the quadratic minimization problem
(13.2):
H 1
min
E LSCM (
u
) ,
with U
(S)= {
u
(S) ,
u
=
u D (
x
)
on
S 0 },
(13.5)
u
U
( S )
where
S 0 is a closed curve of
S
. We assume the following finite expansions for
u
= {
u
,
v
}
:
)= i I u i φ i ( x )+ i J u D ( x i ) φ i ( x )
u h (
x
(13.6)
where I denotes the set of nodes of
S
that do not belong to the Dirichlet boundary, J
denotes the set of nodes of
S
that belong to the Dirichlet boundary and where
φ i are
Search WWH ::




Custom Search