CCTV Video Image Enhancer

This software converts a completely dark image of a bus into an image which identifies the bus. You can see the side livery, make out the brand and the size of the vehicle to identify it. Watch the conversion of the image take place.

---

Closed Caption:

in this example we have a basic
collector optical sensor that has a very
simple night vision feature and were
examining a large vehicle in this case
we're in an urban environment and so the
large vehicle is a bus but the same
principles can apply to any other large
vehicle
so first thing we do is the interlace
the video and now we use the histogram
equalization with the intensity mode to
intensify the light you can see how fast
it is to remove the from the black and
amplify the little signal in the light
that's there
at this point we see that it's a bus and
we want to see something more from the
bus
we actually want too much better clearer
picture because we want to see any see
some details here so we select some
frames
the bus is filmed from the same position
from around here to here
now the next thing we want to do is
stabilize the image will select image
stabilization local stabilization and
we'll take the the the tire as a frame
of reference
here's a result of our stabilize video
you can see an improvement and basically
all the frames follow the same where the
buses in the same position
so now what we want to do is integrate
all the frames together so we'll use
frame averaging and here it runs
even though this is a fairly complex
operation it does run relatively quickly
so we can enhance the frames and stack
frame upon frame so that we reduce the
noise and the problems caused by all the
different values for light and here we
can see the result
so we can clearly see that it's a bus
and we can actually see some writing on
the side and some other details
now we'll use the curves function and
we'll play around with it until we get
the best curve and what the curves does
is it matches that it Maps the input
image values to the output image values
according to the drawing curve the x
coordinate of the curve is mapped to the
court to the corresponding Y value and
what this does is it provides us with a
very quick method of correcting light
color and and intensity of of values
now we can go through with sharpening
make these adjustments and now you can
see from the original very black image
we go down through
and we can see some of the details on
the side of the bus
the UH this would be very helpful if you
want to determine the markings on a
vehicle and whether it's friend or foe
or if they have cargo people that type
of thing and you don't necessarily have
the the ability to use the best type of
electro optical sensor to do with the
job
so you might have to do some post image
processing ore processing in real time
as we're able to do with with our
product as well