
320 Variational Methods in Image Processing
Let us recall the minimization problem when p = 1,
inf
(χ
E
1
,...,χ
E
m
,w)
G
AP S
n
(χ
E
1
, ..., χ
E
m
, w) : χ
E
j
∈ {0, 1}, χ
E
j
∈ BV (Ω),
w ∈ BH(Ω)}, (14.8)
where
G
AP S
n
(χ
E
1
, ..., χ
E
m
, w) =
X
1≤j≤m
µ
j
ˆ
Ω
|Dχ
E
j
|
+
X
1≤i≤n
λ
i
ˆ
Ω
|u
0
− c
i
(χ
P
i
, w) − w|
2
χ
P
i
dx
+ ν
1
ˆ
Ω
|∇w|dx + ν
2
ˆ
Ω
|D
2
w|,
and c
i
are given as before by (14.2). Recall that
´
Ω
|D
2
w| = |D
2
w|(Ω).
Theorem 41 As s ume that Ω ⊂ R
2
is open and bounded, with a C
2
uniform
boundary. Let u
0
∈ L
2
(Ω), µ
j
> 0, j = 1, ..., m, λ
i
> 0, i = 1, ..., n = 2
m
and ν
1
, ν
2
> 0 . Then there are functions χ
E
j
∈ BV (Ω), with χ
E
j
(x) ∈ {0, 1}
dx-a.e. in Ω, and w ∈ BH(Ω) as solutions of problem (14.8).
Proof. T he proof ...