Cross-correlation and image alignment for multi-band IR sensors

We
present
the
development
of
a
cross-­‐correlation
algorithm
for
correlating
objects
in
the
long
wave,
mid
wave
and
short
wave
Infrared
sensor
arrays.
The
goal
is
to
align
the
images
in
the
multi-­‐
sensor
suite
by
correlating
multiple
key
features
in
the
images.
Due
to
the
wavelength
differences,
the
object
appears
very
differently
in
the
sensor
images
even
the
sensors
focus
on
the
same
object.
In
order
to
perform
accurate
correlation
of
the
same
object
in
the
multi-­‐band
images,
we
perform
image
processing
on
the
images
so
that
the
features
of
the
object
become
similar
to
each
other.
Fourier
domain
band
pass
filters
are
used
to
enhance
the
images.
Mexican
Hat
and
Gaussian
Derivative
Wavelets
are
used
to
further
enhance
the
features
of
the
object.
A
Python
based
QT
graphical
user
interface
has
been
implemented
to
carry
out
the
process.
We
show
reliable
results
of
the
cross-­‐correlation
of
the
objects
in
multiple
band
videos.