SeDDaRA
was developed while working with severely degraded images.
However,
we quickly realized that the process can be applied to one-dimensional
signals,
such as sound waves or communications, as well. On this page we
demonstrate
the effectiveness on synthetic data, and on recorded sound waves.
You
can listen to the sound waves on the Sound
page.
Fig1:
The top graph is synthetic data formed using basically a random number
generator.
A known impulse function was applied to the data which produces
our
degraded data. The third graph down is the deconvolution of the
using
the known impulse function. The last graph is the restoration of
the
data using blind deconvolution (impulse function is not known) with
application
of SeDDaRA. The process restores almost as much information as if
the
impulse function were known.
Fig1: This is a computer generated
soundwave that sweeps a
frequency range from 200 to 600 Hz with a level amplitude of 1.32 units.
Fig 2: This sound wave was played through the
computer
speakers and simultaneously recorded by the computer microphone.
As
you can see, the result is not a very good reproduction of the
original.
Run your mouse over the
image to reveal the restored figure.
Fig 3: The signal was processed using our
one-dimensional
SeDDaRA program. Although there is some amplification of noise in
the
higher frequencies, the signal has improved tremendously. It
sounded
better too.
