Re: pixels, bis

[Kennedy McEwen <rkm@kennedym.demon.co.uk>]

In article <B7A41FE6.809F%wirem@home.com>, Wire Moore <wirem@home.com>
writes
>
>I suggest that the term pixel is not abstract in the sense of the term
>point.
>
Neither do I.  Just because a pixel does not possess dimensional
information does not make it a point, just as a square is nor a point,
but does not have dimensions.  in this respect, pixels certainly are
more like squares, triangles, circles etc. in that their size is defined
by the user under certain conditions.
>
>It is appropriate to talk about substantive pixels. I submit we find pixels
>in the structure of camera CCDs, we find pixels on the faces of CRT
>displays, we find pixels in printed output, we find pixels in the data
>representation of images. These are all substantive, true, appropriate and
>worthy uses of the term pixel. These all can be located, measured,
>quantified and compared by any of us. Even within Photoshop, pixels have
>definite extent, and this extent varies depending on what mode the image is
>in, RGB vs. Lab, vs CMYK. Without such extent, a computer could not operate,
>let alone represent images. Without extent, the term pixel would be useless
>because it would refer to nothing. This is not true of the term point.
>
That is completely missing the point! (pun intended)  The term pixel is
neither point nor dimension.  It is a picture element, pure and simple -
the smallest picture element which conveys meaningful information about
the image which contains it.  The image equivalent of the "bit", from
binary digit - the smallest amount of logical information.

Certainly monitors, displays, prints, CCDs etc. contain elements which
are capable of supporting pixels, isolating them from the rest of the
image, identifying them but it is completely wrong to consider them to
be containing pixels.

Lets consider your argument concerning CCDs for example.  You consider
the CCD to contain pixels and that must, by definition, determine the
number of pixels in the image that the CCD can produce - but you would
be completely wrong, both at the upper end of the comparison and the
lower.

Binning of CCD elements, on chip or off, means that the image produced
by the CCD can have substantially fewer pixels than CCD elements, so
there is no match at the lower end.

More importantly, each CCD element has a resolution which extends well
beyond the capability of the array structure to reproduce.  Using a
technique known as microscan, this additional resolution can be
extracted, resulting in many more pixels being produced in the image
than there are CCD elements.  This is a well established fact - I know
because I hold the patents for many of the practical methods of
achieving this.  It has been deployed on the Hubble telescope and on the
imager on the Mars rover.  More recently, my group deeveloped the worlds
highest resolution (confirmed by actual measurements) uncooled thermal
imaging camera by producing images of more than twice as many pixels as
the elements in the sensor array.

Similar, but not identical techniques are implemented in many commercial
systems, such as Epson scanners or Fuji digicams - all produce more
pixels than the elements in the CCD would suggest was possible.  These
are real pixels which can be measured, so they are not simply obtained
by interpolation, although many who do not understand the issues
consider this to be the only description that makes sense to them.

So, there is no match between the number of CCD elements and the number
of pixels in the image at the upper end of the comparison either.  In
short, there is NO relationship between CCD elements - or pixels as you
would wish to call them - and pixels, other than the wish of the
simpleton that all other issues would disappear.

So whilst the term may well be abstract and dimensionless, the issues
caused by its misuse and needless fixation to dimensions are
substantial.

I am quite happy that many people, some highly qualified engineers,
cannot grasp the difference between pixels (picture elements) and CCD
cells or monitor dots.  That is why my sensor designs consistently
outperform my competitors on the resolution stakes.  When the
accountants ask how I can get more pixels in my camera's images than the
sensors used to produce them have I simply refer them to Azimov's first
law - "any technology is indistinguishable from magic to those who do
not understand it".  

>Kennedy's definition of pixel obliterate my reality.

Your failure to understand the abstraction says more about you than the
concept you are attempting to refute.  See previous paragraph!

Breaking the link between sensor and pixel may be your first step to
understanding image reality.

>
>Kennedy and Laurie: if you would put forth a more formal definition of
>pixel, then your points could serve some purpose. I predict that if you do
>provide such a definition, you will find your previous points suddenly
>uninteresting.
>
As I previously posted Wire, there is no need of an additional
definition.  The original definition of "picture element" serves
perfectly well.
-- 
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers
-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.