Graphics Reference
In-Depth Information
Signed deviation color map and distribution
between restored face and gallery face
Face after
curves restoration
Face after
occlusion removal
3.46655
2.73649
2.00643
1.27636
0.00000
Total distribution:
99.89693%
average
-0.91382
-1.64388
-3.10401
-3.83407
3.46000
3.11400
2.76800
2.42200
2.07600
1.73000
1.38400
1.03800
0.69200
0.34600
0.00000
Restoration
Mesh
generation
-2.37395
Total distribution:
98.57543%
average
Absolute deviation color map and distribution
between restored face and gallery face
Neutral face
(Gallery)
Alignment
Restored face
Figure 3.22
Illustration of a face with missing data (after occlusion removal) and its restoration. The
deviation between the restored face and the corresponding neutral face is also illustrated. Copyright
C
1969, IEEE
Algorithm 6
Curves restoration
Input
:
P
: face with missing data, eigenvectors
B
k
j
,
k
=
1
...
K
Output
:
P restored
K : number of curves in each face; n: number of points in a complete curve.
for
k
1
to
K
do
if
quality(c
k
=
←
0)
then
for
i
1
to
n
do
flag
k
[
i
]
←
1;
if
i corresponds to missing part
then
flag
[
i
]
=
=
0;
end
end
q
k
=
SRVF
(
c
k
);
exp
−
1
v
k
=
μ
k
(
q
k
); if
flag
[
i
]
=
1
μ
k
(
N
B
k
q
k
=
exp
j
=
1
c
k
v
k
j
v
k
j
); if
flag
[
i
]
=
1
end
end
3.8.2 Hierarchical Organization of Facial Shapes
One of the main goals for studying shapes of the nose/face region is to conduct biometric
identification where query is often compared to a set of gallery shapes. This comparison