WD said:
Kennedy,
Couldn't an architecture be devised whereby with a CCD gamma compensation
could be done prior to digitisation?
For example:
===== ==================== =====
CCD Analog output--->| PGA |--->| Analog Gamma Comp. |--->| A/D |
===== ==================== =====
The CCD output is essentially analog (although of course discretely
sampled in space), the PGA would be a programmable gain amplifier which
could be used for the 'digitial non-uniformity cal.', This is followed
by an analog circuit which does gamma compensation prior to the A/D
converter.
Wow that takes me back to my youth! ;-)
I held some long since expired patents that addressed essentially this
very approach - albeit for different reasons - and it was put into
production in several systems.
Indeed, prior to the days of high speed, high bit depth ADCs, this was
the *only* way of correcting the CCD output in low contrast situations,
such as I encountered with early infra-red focal plane arrays.
However it doesn't just require a programmable *gain* amplifier, but a
programmable dark current removal too.
These functions are most logically undertaken by storing the offset and
gain data digitally (as at present) and then applying the offset through
a D/A convertor fed to an analogue subtraction circuit to remove the
dark current variation. The result is then passed to the analogue input
of a multiplying DAC, or MDAC and the gain correction data applied to
the digital input, to produce a linear calibrated analogue output which
can then be processed in analogue and digitised as required. In the
70's and 80's DACs and MDACs were available at video speeds with
adequate resolution, whilst fast, economic high resolution ADCs only
became available fairly recently.
When I first proposed this approach, nobody took it seriously because
the world was transitioning to digital technology and the concept of
precision analogue computation was considered to be old fashioned, bulky
and expensive. I had to manufacture some new cameras that were a
fraction of the size, power consumption, weight and cost of the
competition to demonstrate its worth. I remember one of the world's
experts in thermal imaging technology actually asking to look round the
back of our exhibition stand, because he didn't believe the images could
be produced in the cameras on display without a rack of power guzzling
ADCs digitising the data!
Would this accomplish what you are referring to with the PMT analog
gamma comp in drum scanners?
It would, but the performance achieved would not be any better than what
can currently be achieved by digitising the CCD output directly. The
bottleneck in the approach is the gain correction step, which can't be
achieved with any better precision than 16-bits at a sensible speed.
Consequently, the approach has long since become obsolete and we let the
patent lapse.
If your local fire department still uses an old CairnsIris helmet
mounted thermal imaging system for search and rescue, you might be able
to see an imaging sensor that used that principle in operation - if not,
a Google on CairnsIris pulls thousands of links up. At the time it was
first introduced, it was revolutionary, because the thermal contrast in
a scene, particularly in smoke, is extremely low and this required
16-bit precision calibration in real time to extract a useful image from
the low detector signals. This meant the conventional approach would be
much too heavy and consume too much power to fit on a helmet, which was
important when the firefighter had to use both hands to operate
equipment or rescue people. The analogue processing made the whole
concept of a hands free imager that could allow rescuers to see through
smoke right to the fire victim possible.
These days, the detector output is just digitised directly - often on
the focal plane itself sometimes with the dark current corrected in
analogue first, but never the gain any more - and the signal extracted
from the digital data, just like it is with scanners. CairnsIris was
the second last product I designed using this analogue calibration
approach.
