Re: Multi-sampling and "2400x4800 dpi" scanners

In article <43275103.1B9077A7@xxxxxxxxxxxxx>, Gordon Moat <moat@xxxxxxxxxxxxx> writes 
Kennedy McEwen wrote:

In article <4325DEBE.E10868F9@xxxxxxxxxxxxx>, Gordon Moat
<moat@xxxxxxxxxxxxx> writes
>
>Actually, many linear CCDs are 8400 or 10200 cells (pixel sites), though
>divided by three to give each colour Red, Green, and Blue. Kodak have some
>nice White Papers on these.
>
Not generally - colour linear CCDs used in scanners are generally
tri-linear.  Each colour is a separate parallel line of CCDs, they are
not divided.. . . . . . . Now, I am not saying these devices don't exist, but
I would like some
pointer as to where you are getting this information from since it is
not from the Kodak or Dalsa sites you reference, and more likely to be a
misunderstanding on your part.  Whilst there may well be colour
interleaved linear CCDs these are certainly not used on any commercial
scanners that I am aware of.


Okay, maybe I should have stated that better. So I will give you one to find
and read about. That is the Kodak KLI-10203 Imaging Sensor. It is correctly
termed a 3 x 10200 imager, so I apologize for not being more thorough in my
description of it.

The KLI-10203 is a tri-linear CCD (check the FIRST line of the data ***!) - *each* of the lines is 10200 cells long and *each* of the lines is a separate colour - no interleaving. So, contrary to your claim that this could only resolve 3400ppi, because it has 3 colours in each line, it can 5100cy/length. Without optical scaling it resolves 3600ppi, with optical scaling (as would be used in a scanner application) this can be set to match whatever the scanner width is - on the 8.5in flatbed scanner configuration that the OP is referencing, it would produce around 1200ppi. 

>So in theory an 8400 element linear CCD should be able to resolve 2800
>dpi, and a 10200 element CCD should be able to do 3400 dpi.

A colour CCD with a total of 8400 elements would only be capable of
resolving 2800 colour samples across the A4 page - somewhat better than
300ppi - whilst your 10200 element colour CCD would only be capable of
400ppi!  The real requirements for flatbed scanners are *much* higher
than these!


If you could figure out what scanner uses the KLI-10203, then you might be
surprised at your statements.

I don't think so, mainly since the statement is based on *YOUR* figures that the 10200pixel CCD is only capable of 3400ppi! Perhaps you see now why it was ridiculous? And before you wriggle further - KODAK DON'T MAKE A 3400PIXEL LONG TRILINEAR (10200 TOTAL CELLS) CCD AND NEVER HAVE!!

An A4 flatbed scanner, as the type under discussion in this thread, means a scan width of at approximately 8.5"; 10200 pixels across that distance yields exactly 1200ppi - no division by three because the colours are on three separate lines of 10200 cells *each*, not interleaved on a single line as you suggested.

Just to give you a hint, it is only available in
a few high end products. The lowest spec (and lowest cost) of those does 3200
dpi true resolution. That is across the entire bed, and not just down the
middle.

Not across the full A4 width o a single pass it isn't. To achieve 3200ppi resolution requires a scan width of no greater than 3.2" - around a third of the width of the flatbed under discussion! 

An A4 scanner with a 1200ppi capability has a tri-linear CCD with round
10,500 cells in *each* line ie. a total of more than 31,000 cells.  A
4800ppi full page scanner requires a CCD with more than 42000 cells in
each line, a total of over 125,000 cells.


Okay, just to through out some numbers, and then you can do calculations, or
whatever. Using the KLI-10203 again, the cell sites are 7 μm square pixels.
There are 3 rows of 10200 cells each, so 30600 total cells. Row spacing is 154
μm centre to centre. There is no sideways offset of cells in each row, and the
spacing allows a processing timing gap of 22 lines.

And how much of that determines the ppi of the final application? Hint - nothing, but now we know you can read a data ***!

So where does *your* figure of 3400ppi limitation for this particular device come from - apart from your initial misreading of the data?


Dalsa bought out the Philips imaging chip business, though they kept some
engineers and other workers. Is it still possible to buy imaging chips directly
from Philips?

Certainly was the last time I tried, which I believe was earlier this year although time flies.

Anyway, they do have some nice information on chips on their
website.

They do, but *none* of them are linear arrays and making inferences from the limitations of 2-D arrays, particularly colour arrays, on linear devices is misleading at best and completely deceptive at worst. For example, DALSA's biggest array is only 5344 pixels along the largest axis - but you wouldn't interpret that as state of the art for a linear array!


Interleaved colours (by Bayer masking) is common on two dimensional CCDs
(indeed, Bayer was a Kodak employee!) but this is unnecessary in linear
devices.  I suspect that you are confusing the two.


Okay, to be more specific, each row on a linear CCD has a colour filter over
it.

Precisely - but that isn't what you wrote last time! You stated that the 3 colours resulted in a resolution of only one third of the number of pixels in the line.

<snip the millenium prize for rewording the previous post!> 

Okay, so I don't recall mentioning interleaving, but interpolation was
mentioned, though only for upsizing or downsizing to change resolution. The OP
wants to use what he thinks might be extra resolution in one dimension of the
specifications for his scanner.

No he doesn't - or at least that isn't what he has asked about. He is interested in using available samples in two axes that do not provide as much resolution as he would like as a means of achieving improved signal to noise at a lower resolution.

The CCD in his case is similar to the NEC uPD8880 device, a trilinear array with 21360 cells in each colour, capable of producing 2400ppi across an A4 platform. Each of the colour lines comprises two rows of 10,680 cells capable of reproducing 1200ppi on the flatbed, but offset by half a pixel pitch to create a 2400ppi sample density. In addition, the scanner motor is capable of moving the scan head in 4800ppi steps, further oversampling the original pixels. He is interested in using these oversamples optimally for signal to noise improvement at 2400ppi and possibly as low as 1200ppi rather than have some of their information being used to achieve resolution which is already compromised by the optical system of the scanner.

An exception to colour filtering is in many Nikon film scanners, since they use
coloured LEDs as a light source. I would suspect those are Sony imaging chips
in those Nikon scanners.


You would be wrong.

While many do like the LED approach, it is interesting
to note that is not done in any high end scanning systems.

Wrong again! It is exactly the process used in high end film scanner systems - the difference being that the LEDs are replaced with colour lasers to achieve a higher intensity and thus a faster throughput.

I doubt it is some
patent issue, and more likely that a single light source provides a more
predictable scanning operation in regards to colour accuracy over the life of
the scanner.

Anyway, I apologize for not being more clear, a 10200 linear CCD should be
correctly termed a 3 x 10200 element linear CCD. Regardless the resolution is
still limited by the physical size of the cell site, the scanner optics, and
the accuracy of movement of the imaging components within the scanner. A linear
image sensor with a single array of 1000 photosites of pitch 10 μm would have a
resolution of 2540 dpi (1000 / (1000 x .01 mm x 1"/25.4mm)). If that sensor
were used in an optical system to image an 8" wide document, then the
resolution in the document plane would be 125 dpi (1000 pixels / 8"). If we
consider the 7 μm cell size for the KLI-10203, for example, then we can
estimate for that imager.

You don't need to go round the houses - the calculation is trivial. An 8.5in scan width with 10200 cells per line (no matter what the optical system or the cell size or pitch is) results in 10200/8.5 = 1200ppi.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
.