|
|
Hi,
I can't resist making a comment. Having my two bob's worth.
As I see it, a pixel in an image has the following properties.
A position, based on some coordinate system, which in the case of ImageJ and
many image processing and analysis programs, has an origin at the top left
of the image, with the column index running from left to right (increasing)
and the row index running from top to bottom (increasing). Just like the
writing on the pages of a book in English. (assuming the pixels are square).
And the pixels have a physical size, because images are usually of real
objects. So some spatial scale is involved.
Each pixel is a sample of the intensity of light from some object, which is
produced by some physical transformation process in which light (usually,
but could be other radiation, or signal source) is converted into a signal
which ends up in computer by various processes, none of which are simple,
and some which appear almost miraculous.
The degree to which the pixel in the sensor samples the intensity of the
measured signal, evenly over its areal extent is a fact that may be ignored
by many, knowing about this uncertainty is important.
If some information theory has been employed then maybe the pixel size will
be such that the detail in the source is spread out over the pixels in
accordance with the Nyquist limit, i.e. at least two samples or better in
each direction for each element of of the source data. For example, in the
case of an optical system the diameter of the diffraction spot (or
resolution limit) might be used as a guide to setting up the system.
So in summary a pixel will usually be a representation of the intensity or
brightness of part of an object, which indicates how bright or dark that
part of the object/sample is. How accurate the representation is may be
determined by intensity calibration. And testing an object using a range of
different pixel sizes on the image (if possible).
A pixel may be mathematical point sample, but it usually represents an area
sample, and covers a finite area.
If there are errors in this small essay, please tell me.
Noel
--
Noel Goldsmith
Air Vehicles Division
DSTO
506 Lorimer Street
Port Melbourne
Vic 3207
AUSTRALIA
Phone 96267527
Fax 96267718
|
Locked
