|
|
Satellite Technology
In this world of rapid technological advancement few scientific issues
still capture the public interest, although satellite imagery does
seem to be an exception.
In fact the general public’s
perception appears to be that satellite imaging is far in advance of the current reality. Satellites
that identify and follow individuals through police chases, that are available anywhere in the
world 24 hours a day 7 days a week at the whim of the operator and that can produce colour
pictures through cloud cover are all, at this time, fiction.
What then is the reality? How does an
orbiting imager some 800 km above the surface of the earth collect the pictures that we are so
familiar with, and what are the real capabilities and limitations of this technology?
"One pixel at a time"
We need to start with how the satellite acquires its images. Unlike a camera which acquires a
complete frame of data in a single exposure most satellites acquire their images one pixel or one
line at a time. Each pixel in a satellite image is composed of a number of “bands” of data where
each “band” is sensitive to a particular portion of the electromagnetic spectrum. This data is
recorded digitally as a discrete intensity value for each pixel in each band and as digital data can
be transmitted electronically to a processing system that can display the imagery as a picture.
The following illustration is a simple and non specific diagram indicating the general principals
behind satellite image acquisition, transmission and processing.
"Only a few satellites capture color across the entire visible spectrum"
These bands may be in the visible portion of the electromagnetic spectrum or not. Much of the
value of satellite imaging is in its acquisition of data outside the visible portion of the spectrum.
As a consequence, only a few satellites capture data across the entire visible spectrum and those
that do are acquiring data that is visibly quite different than what you and I are normally used to
seeing. The atmosphere provides a blanket that absorbs and disperses much of the light energy as
it passes through to earth and the light that is reflected back to the satellite has to pass, a second time, through this same blanket.
The result is an image that is devoid of much of the natural
colour that we are so used to from our vantage point at the surface of the earth. This twofold
problem (not collecting the entire visible spectrum and the atmospheric loss or absorption of
light) can be seen below:
Techniques for adding color
In order to compensate for this difference techniques are applied using some of the non visible
information to correct for these absorption factors.
Here is a before and after example:
This is only the beginning of the problems faced by the image processor
in dealing with satellite imagery. Resolution is the next factor that
needs to be considered. 
High image resolution and spectral information
The images above are very
high resolution images and are good for assessing local areas. However if you are concerned
about regional data or larger areas you need to acquire imagery at a lower resolution.
This is true
for two main reasons; - lower resolution imagery taken over larger areas tends to produce better
spectral information for that resolution, and
- the cost is considerably lower than covering the
same area with high resolution data
When one zooms into a satellite image beyond its natural
resolution the image starts to “pixelate” as can be seen in the image to the right
How Zoomify technology works
So, in order to generate a zoom sequence that would take you from space to a local street you
must pass through a number of layers of imagery at different resolutions. In fact you would need
to transit through a total of at least four (4) different layers to accomplish this task properly. The
difficulty comes in the fact that the colours at the different resolutions are not consistent and in
order to generate a continuous zoom that looks real it is important to balance the colours between
layers so that the zoom sequence is seamless. The following sequences illustrate this problem:
Clearly the colours between the 1 Km. Data and the 100 m. Imagery
are not consistent. Nor is the colour balance between the 100 m data
and the 5 m data. Finally there is a colour shift seen between the
5 m data and the 1 m imagery (although the colour balance between
these two images is better than any of the other pairs). How then
can we reduce this effect? The following sequence shows imagery that
has been adjusted to minimize these effects while still maintaining
much of the excitement of the lower resolution data.
The animated sequence that follows shows how realistic such a sequence can look.
The world is a large place and the cost in time and money to explore our home is beyond the
capabilities of most people. The application of this technology now allows us to reduce the cost
of exploration and expand the territories for those armchair explorers who care to take advantage
of the possibilities.
|
Suggest
a Print
