how does image scanner calibrate itself?

  • Thread starter Thread starter bunnypartz
  • Start date Start date
B

bunnypartz

Hi, I posted earlier that I was getting a couple streaks on the scan,
vertical streaks ... I now believe it's a calibration problem, or perhaps
it is, I have reason to believe it from what I've read so far.

How does a scanner calibrate itself? I mean, where is the calibration
target, and what part is looking at what part ... I just have no idea. If u
know of a site that explains it, I'd be eternally grateful, or somewhat
grateful .. or if u wanna take the time to explain it. There's also
something called an image guide, which ... is related to calibration??
Anywayz I'll continute to search for info on the net in the meantime.

THX
 
Hi, I posted earlier that I was getting a couple streaks on the scan,
vertical streaks ... I now believe it's a calibration problem, or perhaps
it is, I have reason to believe it from what I've read so far.

How does a scanner calibrate itself? I mean, where is the calibration
target, and what part is looking at what part ... I just have no idea. If u
know of a site that explains it, I'd be eternally grateful, or somewhat
grateful .. or if u wanna take the time to explain it. There's also
something called an image guide, which ... is related to calibration??
Anywayz I'll continute to search for info on the net in the meantime.
Click to expand...


Vertical stripes can also be the CCD, cells either blocked by dust, or the
CCD can be defective. It can also be dust or a haze film on the internal
mirrors that reflect the image onto the CCD. Thin multicolor vertical
stripes suggest random dust on CCD or mirror, blocking the image path. Very
wide bands suggest CCD chip defects or a film on the mirrors.

Flatbeds calibrate white balance via a reflective strip on the bottom side of
the top case cover, just over the parked carriage, before the carriage moves
out onto the glass area to scan. It can be an actual white paper strip, it
can be just the plastic cover, or it can be the bottom side of the ruler
scale at the edge of the glass at the carriage end. Glass cleaning fluid can
run under the glass edge discoloring this calibration strip in bands.
 
bunnypartz said:
Hi, I posted earlier that I was getting a couple streaks on the scan,
vertical streaks ... I now believe it's a calibration problem, or perhaps
it is, I have reason to believe it from what I've read so far.

How does a scanner calibrate itself? I mean, where is the calibration
target, and what part is looking at what part ... I just have no idea. If u
know of a site that explains it, I'd be eternally grateful, or somewhat
grateful .. or if u wanna take the time to explain it. There's also
something called an image guide, which ... is related to calibration??
Anywayz I'll continute to search for info on the net in the meantime.
Click to expand...
In addition to Wayne's comment, the calibration should have at least two
sections, not just the white response, but a black response as well.
This is usually achieved by scanning a non-reflective or
non-transmissive area.

The output viewing the black level *should* be zero, but it rarely is
due to dark current which flows in the chip and varies with operating
temperature (hence the need to calibrate before the scan. Any output
produced by a CCD cell when viewing the black reference is subtracted
from the output produced when scanning the image to compensate for this
dark current.

The output when viewing the white level *should* be uniform, but not
necessarily peak white. The only variations that should be present are
response and illumination differences across the CCD, and these are
typically quite stable with operating temperature. So the ratio of the
output from each CCD cell to the average of all cells when viewing the
white reference is then multiplied with the output produced when
scanning the image.

In combination, these two operations are just a simple linear correction
for each CCD cell output. In other words, Y = m X + c, Y being the
desired response, X the actual response, m the white correction term and
c the black correction term.

So for the CCD cell number i, the correction applied during the scan is
just Pixel(i) = White(Ave) x (CCD(i) - Black(i)) / White(i)
where:
Black(i) is the output produced by CCD cell i viewing the black surface
White(i) is the output produced by CCD cell i viewing the white surface
White(ave) is the average output produced by all of the CCD cells when
viewing the white surface
CCD(i) is the output of CCD cell i when viewing a particular part of the
image
Pixel(i) is the expected output of an ideal CCD cell i when viewing that
part of the image.

The process of obtaining these black and white reference responses for
all of the cells in the CCD is the calibration process.
 
carriage end. Glass cleaning fluid can run under the glass edge
discoloring this calibration strip in bands.
Click to expand...

Hi ... thx for yer messages, u and Kennedy, I saved both yer messages ...
I think yer on to something, the calibration strip is on the outside part
of the glass facing in, near the top ... a white one and a black one,
joined length-ways, the black one wider. Shows u what I know, never having
taken the glass off before, I didn't even know theses strips existed there!

.... so anyhow I looked closely, and the first thing I see is a patch of
water about a centimeter wide across the white and black ... water seeped
right in there ...

Makes sense cuz I had noticed the ghosty stripe changed size when I tried
cleaning the glass a second time: I musta contributed more water and made
the patch bigger!!

THX! That I should be able to fix this is incredible ... considering how
clumsy I am ... already a little plastic piece with two pins got dislodged
and I had to guess where it went ...
 
Update on it ... I patted the strip dry, and put the whole thing back
together, and the results were better ... I can't say it's perfect, there's
still some lines now, more or less in the same area, much more thin this
time, and fainter ... I dunno, I'll try cleaning it again, but maybe the
strip was stained slightly ... or, this is what I'm wondering about: or ...
or when I put the calibration strip back, and it isn't exactly in the same
location, maybe off ever so slightly from it's orginal position, could that
effect the calibration ... I mean if it's just a tiny bit off where it was,
like a millimeter ... well I'll find out anyway cuz I plan to adjust it ...

I see two tiny tiny spots on the calibration strip, not sure if they're
glue spots, or what, they kinda off from the calibration area, but maybe
they close enough, and they are near where the water was! So I'll try to
remove em assuming they ain't bare spots.

I was disappointed tho after my initial glee to see these very thin lines
still appear ... but it's gone from terrible -> I can get by with it. And
it's great to know what was causing it.

THX again .
 
bunnypartz said:
Update on it ... I patted the strip dry, and put the whole thing back
together, and the results were better ... I can't say it's perfect, there's
still some lines now, more or less in the same area, much more thin this
time, and fainter ... I dunno, I'll try cleaning it again, but maybe the
strip was stained slightly ... or, this is what I'm wondering about: or ...
or when I put the calibration strip back, and it isn't exactly in the same
location, maybe off ever so slightly from it's orginal position, could that
effect the calibration ... I mean if it's just a tiny bit off where it was,
like a millimeter ... well I'll find out anyway cuz I plan to adjust it ...

I see two tiny tiny spots on the calibration strip, not sure if they're
glue spots, or what, they kinda off from the calibration area, but maybe
they close enough, and they are near where the water was! So I'll try to
remove em assuming they ain't bare spots.

I was disappointed tho after my initial glee to see these very thin lines
still appear ... but it's gone from terrible -> I can get by with it. And
it's great to know what was causing it.
Click to expand...
Anything that you can see on the calibration area, even specs of dust,
will give rise to stripes in the image. If you are really stuck and the
surface is stained irremovably, then you might be able to get hold of
some black and white smooth plastic strips that could be stuck over the
original calibration surfaces. One suggestion I have heard was to use a
roll of that iron-on edging for melamine boarding. Most hardware stores
have it in a white finish. Obviously stick it in place, don't iron it
onto the scanner! The white stuff should work almost as is, while some
flat matt black Krylon spray paint will create a good black reference.
Don't spray the strip anywhere near the scanner obviously. Being cheap,
and available in fairly long rolls, you can have several attempts to get
it right, and can cut it to the exact size and finish quite separately
from fitting it to the scanner.
 
Hi ... thx, I'll look for specks of dust, I haven't really inspected the
white as closely as the black, cuz with the white it is difficult to see
anything. After awhile, my eyes feel like they're gonna explode.

If it is stained, would acrylic paint work at all??! I'm a painter and got
lots of paints, I tend to see everything as a painting matter... ha ...
I'd be afraid to try it cuz if it doesn't work then it's finished.
 
bunnypartz said:
Hi ... thx, I'll look for specks of dust, I haven't really inspected the
white as closely as the black, cuz with the white it is difficult to see
anything. After awhile, my eyes feel like they're gonna explode.

If it is stained, would acrylic paint work at all??! I'm a painter and got
lots of paints, I tend to see everything as a painting matter... ha ...
I'd be afraid to try it cuz if it doesn't work then it's finished.
Click to expand...

If you are confident that you can create a good, flat, clean finish then
paint will work as well as anything - but it is a lot more permanent, so
if you screw up you can't rework. I would rather paint some flat
disposable material I can overlay on the calibration area than try to
paint it directly.
 
permanent, so if you screw up you can't rework. I would rather paint
some flat disposable material I can overlay on the calibration area
than try to paint it directly.
Click to expand...

Good point! Very good point. Well ... I basically got the residual streaks
out ... the final very thin streaking seems to have been caused by dust on
either side of the glass in front of the strip ... so dust on the glass in
front of the strip added some streaking ...

So now the strip is clean and the glass is clean, and I got decent images.
Now I realize I never had to take the scanner apart, cuz the strip is on
the outside of the glass facing inward ... but if I didn't take it apart,
I'd never have known this thing was a calibration strip, and that water was
on it ... unless I was able to make the connection from what ppl said here,
without visually seeing it.
 
Back
Top