<s>
YaDICs	B-Language
is	O
a	O
program	O
written	O
to	O
perform	O
digital	O
image	O
correlation	O
on	O
2D	O
and	O
3D	O
tomographic	O
images	O
.	O
</s>
<s>
Before	O
the	O
appearance	O
of	O
such	O
methods	O
,	O
researchers	O
usually	O
used	O
strain	B-Algorithm
gauges	I-Algorithm
to	O
measure	O
the	O
mechanical	O
state	O
of	O
the	O
material	O
but	O
strain	B-Algorithm
gauges	I-Algorithm
only	O
measure	O
the	O
strain	O
on	O
a	O
point	O
and	O
do	O
n't	O
allow	O
to	O
understand	O
material	O
with	O
an	O
heterogeneous	O
behavior	O
.	O
</s>
<s>
There	O
is	O
another	O
process	O
called	O
image	B-Algorithm
registration	I-Algorithm
using	O
the	O
same	O
algorithms	O
(	O
on	O
monomodal	O
images	O
)	O
but	O
where	O
the	O
goal	O
is	O
to	O
register	O
images	O
and	O
thereby	O
identifying	O
the	O
displacement	O
field	O
is	O
just	O
a	O
side	O
effect	O
.	O
</s>
<s>
YaDICs	B-Language
uses	O
the	O
general	O
principle	O
of	O
image	B-Algorithm
registration	I-Algorithm
with	O
a	O
particular	O
attention	O
to	O
the	O
displacement	O
fields	O
basis	O
.	O
</s>
<s>
YaDICs	B-Language
can	O
be	O
explained	O
using	O
the	O
classical	O
image	B-Algorithm
registration	I-Algorithm
framework	O
:	O
</s>
<s>
The	O
common	O
idea	O
of	O
image	B-Algorithm
registration	I-Algorithm
and	O
digital	O
image	O
correlation	O
is	O
to	O
find	O
the	O
transformation	O
between	O
a	O
fixed	O
image	O
and	O
a	O
moving	O
one	O
for	O
a	O
given	O
metric	O
using	O
an	O
optimization	O
scheme	O
.	O
</s>
<s>
While	O
there	O
are	O
many	O
methods	O
to	O
achieve	O
such	O
a	O
goal	O
,	O
Yadics	B-Language
focuses	O
on	O
registering	O
images	O
with	O
the	O
same	O
modality	O
.	O
</s>
<s>
In	O
order	O
to	O
get	O
sufficient	O
performances	O
OpenMP	B-Application
parallelism	O
is	O
used	O
and	O
data	O
are	O
not	O
globally	O
stored	O
in	O
memory	O
.	O
</s>
<s>
Contrary	O
to	O
image	B-Algorithm
registration	I-Algorithm
,	O
Digital	O
Image	O
Correlation	O
targets	O
the	O
transformation	O
,	O
one	O
wants	O
to	O
extracted	O
the	O
most	O
accurate	O
transformation	O
from	O
the	O
two	O
images	O
and	O
not	O
just	O
match	O
the	O
images	O
.	O
</s>
<s>
Yadics	B-Language
uses	O
the	O
whole	O
image	O
as	O
a	O
sampling	O
grid	O
:	O
it	O
is	O
thus	O
a	O
total	O
sampling	O
.	O
</s>
<s>
It	O
is	O
possible	O
to	O
choose	O
between	O
bilinear	O
interpolation	O
and	O
bicubic	B-Algorithm
interpolation	I-Algorithm
for	O
the	O
grey	O
level	O
evaluation	O
at	O
non	O
integer	O
coordinates	O
.	O
</s>
<s>
The	O
bi-cubic	B-Algorithm
interpolation	I-Algorithm
is	O
the	O
recommended	O
one	O
.	O
</s>
<s>
The	O
SSD	O
is	O
also	O
known	O
as	O
mean	B-Algorithm
squared	I-Algorithm
error	I-Algorithm
.	O
</s>
<s>
This	O
metric	O
is	O
the	O
main	O
one	O
used	O
in	O
the	O
YaDICs	B-Language
as	O
it	O
works	O
well	O
with	O
same	O
modality	O
images	O
.	O
</s>
<s>
This	O
metric	O
is	O
only	O
used	O
to	O
find	O
local	O
translation	O
in	O
Yadics	B-Language
.	O
</s>
<s>
This	O
is	O
commonly	O
used	O
in	O
Image	B-Algorithm
Registration	I-Algorithm
by	O
the	O
use	O
of	O
B-Spline	B-Algorithm
functions	O
and	O
in	O
solid	O
mechanics	O
with	O
finite	O
element	O
basis	O
.	O
</s>
<s>
First-order	O
quadrangular	O
finite	O
elements	O
Q4P1	O
are	O
used	O
in	O
Yadics	B-Language
.	O
</s>
<s>
The	O
YaDICs	B-Language
optimization	O
process	O
follows	O
a	O
gradient	O
descent	O
scheme	O
.	O
</s>
<s>
BFGS	O
,	O
conjugate	O
gradient	O
,	O
stochastic	O
gradient	O
)	O
but	O
as	O
steepest	O
gradient	O
and	O
Gauss-Newton	O
are	O
the	O
only	O
ones	O
implemented	O
in	O
Yadics	B-Language
these	O
methods	O
are	O
not	O
discussed	O
here	O
.	O
</s>
<s>
Pyramidal	O
process	O
used	O
in	O
Yadics	B-Language
(	O
and	O
ITK	O
)	O
.	O
</s>
<s>
We	O
use	O
this	O
strategy	O
in	O
Yadics	B-Language
and	O
we	O
add	O
the	O
median	O
filter	O
as	O
it	O
is	O
massively	O
used	O
in	O
PIV	O
.	O
</s>
