Double the resolution with piezo electronics?

Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

Are you spamming up or trolling us? You need an account to access the
article.

A mechanical shift and 2 exposure doesn't make sense for a lot of
things, esp. things than move. For landscapes, just take 2 pics or 4
pics and merge them together for a panorama.

Rich wrote:
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

In article .com, Rich
writes
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

It has been used in commercial cameras but, as far as I know, no
consumer cameras. In this distinction, military cameras are commercial
products since they are manufactured, distributed and sold through a
commercial network just like any other product.

The technique is known as "microscan" and, working for the company which
holds most of the patents associated with its implementation and as
named inventor on most of those, I would also like to know if anyone is
using it commercially, but probably know most users already.

The use of piezoelectric actuators is only one option - the image simply
needs to move by a fixed and controlled amount between time samples. For
example, a rotating prism or tilted window in the optical path can
achieve the same displacement if synchronised with the sensor. Piezos
on the sensor are the easiest to conceptualise.

The technique is also used in scanners - the early Canon "VAROS" system
used two scans that were opto-mechanically offset by half a pixel, but
was superceded by Epson's "HyperCCD" which simplified the mechanics,
thus reducing cost, by using two linear CCDs offset by half a pixel
pitch. The Canon system infringed one of my patents and they stopped
using it, the Epson approach doesn't and most scanner manufacturers use
a derivative of that now.

It is, however, a gross simplification to say that it "doubles"
resolution - that is only achieved under very specific circumstances.
What it does is double the sampling density, however the system input
MTF remains unchanged. So, *IF* the system was significantly
undersampled - which would cause serious aliasing - then it would double
resolution. If the system is marginally undersampled - some aliasing
occurs but not seriously - then resolution is improved, but may not be
doubled. If the system is not undersampled - no aliasing ever occurs -
then it will not improve resolution at all.

The basic raw CCD or CMOS sensor is undersampled by default, which is
what causes most of the confusion. For example, a 100% fill factor
sensor will have a pixel geometric MTF which falls to zero at the
sampling frequency, however an array of such pixels can only
unambiguously resolve up to half of that spatial frequency.
Consequently, a raw, 100% fill factor CCD will seriously alias images.
Microscan can double the sampling density and thus, in principle, double
the resolution. In practice, however, the pixel geometric MTF is not
the only limitation on the front end camera MTF - for example the lens
has an MTF as does the AA filter, amongst other issues such as camera
motion and shake. These MTFs may be sufficient to limit the combined
lens/AA/sensor MTF to near zero above the Nyquist limit and thus prevent
any increase in resolution from being realised in practice. For
example, early implementations of the Epson HyperCCD were used in
scanners with very poor optics and the claimed resolution was never
achieved and only marginal over the basic single line CCD. More recent
implementations are much better although none so far quite reach double
resolution.

I have designed and built military cameras with 2x2, 3x3 and even 4x4
microscan implementations, providing potential doubling, tripling and
quadrupling of the basic sensor resolution. Each increase in microscan
"order" places even tougher requirements on the performance of the
remainder of the sensor, in particular the optics, and there is little
benefit to going higher than 2x2 unless the basic fill factor of the
sensor is quite low.

The technique is of limited value in still cameras due to potential
subject motion between sub-frames, however there are some very simple
techniques to perfectly compensate for this on video sensors.
Consequently the technique has mainly been implemented on very high
performance video systems and static image scanners.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)

On Fri, 15 Dec 2006 02:47:16 +0000, Kennedy McEwen wrote:

In article .com, Rich
writes
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

It has been used in commercial cameras but, as far as I know, no
consumer cameras. In this distinction, military cameras are commercial
products since they are manufactured, distributed and sold through a
commercial network just like any other product.

The technique is known as "microscan" and, working for the company which
holds most of the patents associated with its implementation and as
named inventor on most of those, I would also like to know if anyone is
using it commercially, but probably know most users already.

The use of piezoelectric actuators is only one option - the image simply
needs to move by a fixed and controlled amount between time samples. For
example, a rotating prism or tilted window in the optical path can
achieve the same displacement if synchronised with the sensor. Piezos
on the sensor are the easiest to conceptualise.

The technique is also used in scanners - the early Canon "VAROS" system
used two scans that were opto-mechanically offset by half a pixel, but
was superceded by Epson's "HyperCCD" which simplified the mechanics,
thus reducing cost, by using two linear CCDs offset by half a pixel
pitch. The Canon system infringed one of my patents and they stopped
using it, the Epson approach doesn't and most scanner manufacturers use
a derivative of that now.

It is, however, a gross simplification to say that it "doubles"
resolution - that is only achieved under very specific circumstances.
What it does is double the sampling density, however the system input
MTF remains unchanged. So, *IF* the system was significantly
undersampled - which would cause serious aliasing - then it would double
resolution. If the system is marginally undersampled - some aliasing
occurs but not seriously - then resolution is improved, but may not be
doubled. If the system is not undersampled - no aliasing ever occurs -
then it will not improve resolution at all.

The basic raw CCD or CMOS sensor is undersampled by default, which is
what causes most of the confusion. For example, a 100% fill factor
sensor will have a pixel geometric MTF which falls to zero at the
sampling frequency, however an array of such pixels can only
unambiguously resolve up to half of that spatial frequency.
Consequently, a raw, 100% fill factor CCD will seriously alias images.
Microscan can double the sampling density and thus, in principle, double
the resolution. In practice, however, the pixel geometric MTF is not
the only limitation on the front end camera MTF - for example the lens
has an MTF as does the AA filter, amongst other issues such as camera
motion and shake. These MTFs may be sufficient to limit the combined
lens/AA/sensor MTF to near zero above the Nyquist limit and thus prevent
any increase in resolution from being realised in practice. For
example, early implementations of the Epson HyperCCD were used in
scanners with very poor optics and the claimed resolution was never
achieved and only marginal over the basic single line CCD. More recent
implementations are much better although none so far quite reach double
resolution.

I have designed and built military cameras with 2x2, 3x3 and even 4x4
microscan implementations, providing potential doubling, tripling and
quadrupling of the basic sensor resolution. Each increase in microscan
"order" places even tougher requirements on the performance of the
remainder of the sensor, in particular the optics, and there is little
benefit to going higher than 2x2 unless the basic fill factor of the
sensor is quite low.

The technique is of limited value in still cameras due to potential
subject motion between sub-frames, however there are some very simple
techniques to perfectly compensate for this on video sensors.
Consequently the technique has mainly been implemented on very high
performance video systems and static image scanners.

It occurs to me that the cost of implementing this on a camera with
moving-sensor image stabilization would be essentially zero other than the
cost of developing the code--the mechanism to move the sensor is already
present.

Am I in error on this?

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

In article , J. Clarke
writes

It occurs to me that the cost of implementing this on a camera with
moving-sensor image stabilization would be essentially zero other than the
cost of developing the code--the mechanism to move the sensor is already
present.

Probably - it depends on the sensor mechanism drive details. On one
system I designed I used another component that the camera required for
quite a different reason to implement the image motion, thus getting the
resolution gain almost for free. So what you suggest is certainly
feasible in principle.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)

In article om, Pat
writes

Rich wrote:
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

Are you spamming up or trolling us? You need an account to access the
article.

It's free.

A mechanical shift and 2 exposure doesn't make sense for a lot of
things, esp. things than move.

How many "fine documents" move?

For landscapes, just take 2 pics or 4
pics and merge them together for a panorama.

That is a variation of the principle.

--
Ian G8ILZ

? "Rich" ?????? ??? ??????
oups.com...

Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx




It's a technique used in astroCCD imaging, called "drizzle", among other
names.

Kennedy McEwen writes:

In article .com,
Rich writes
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

It has been used in commercial cameras but, as far as I know, no
consumer cameras. In this distinction, military cameras are
commercial products since they are manufactured, distributed and
sold through a commercial network just like any other product.

Definately not comercial, first used in the HST I think, using the FGS
to do sub pixel offsets rather than piezo.

--
Paul Repacholi 1 Crescent Rd.,
+61 (08) 9257-1001 Kalamunda.
West Australia 6076
comp.os.vms,- The Older, Grumpier Slashdot
Raw, Cooked or Well-done, it's all half baked.
EPIC, The Architecture of the future, always has been, always will be.

Many fine docs "move" if you're not using a tripod. The better
question is:
How many people buy a camera just to take pictures of fine documents?

Prometheus wrote:
In article om, Pat
writes

Rich wrote:
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

Are you spamming up or trolling us? You need an account to access the
article.

It's free.

A mechanical shift and 2 exposure doesn't make sense for a lot of
things, esp. things than move.

How many "fine documents" move?

For landscapes, just take 2 pics or 4
pics and merge them together for a panorama.

That is a variation of the principle.

--
Ian G8ILZ

In article om, Pat
writes
Prometheus wrote:
In article om, Pat
writes

Rich wrote:
Note this passage from the document listed below. Has this been done
in any commercial camera?

"Resolution can be doubled by mechanically shifting a sensor a half
pitch by a built-in piezoelectric actuator between the first and second
exposures and combining the images. This is useful in capturing an
image of a fine document."

http://www.photonics.com/content/han...ors/82218.aspx

Are you spamming up or trolling us? You need an account to access the
article.

It's free.

A mechanical shift and 2 exposure doesn't make sense for a lot of
things, esp. things than move.

How many "fine documents" move?

For landscapes, just take 2 pics or 4
pics and merge them together for a panorama.

That is a variation of the principle.

Many fine docs "move" if you're not using a tripod. The better
question is:
How many people buy a camera just to take pictures of fine documents?

Archivists, restorers, etc.
--
Ian G8ILZ