Prev: Standalone Parallel Computing
Next: Problem with Move Camera Forward/Back using Camera Toolbar
From: Sami Oueslati on 10 May 2010 16:37
Hi,
Please how to know the size of 1 pixel in cm (or inches) in an image?? Must I know the resolution of my screen?
Thanks.
From: ImageAnalyst on 10 May 2010 18:14
No, the screen has nothing to do with it. Your image could take up
the same, say, 1024 pixels across your screen but if it's an image of
a cell in a microscope or a galaxy from a telescope they'll have
different real world sizes. You need some calibration target in the
plane of the scene. This can be the image or a ruler, or a micron bar
that the microscope puts on there, or a known distance or size of some
known object in your scene such as the angular subtense between two
stars. What do you have in your scene that has a known size?
From: Walter Roberson on 10 May 2010 19:23
Sami Oueslati wrote:

> Please how to know the size of 1 pixel in cm (or inches) in an image??
> Must I know the resolution of my screen?

In addition to the comments made about needing to know a reference value:

In the case where you generate the image and it does not correspond to any
real-world object, and you want to know how big a pixel renders as (e.g., you
might be wanting to calculate how many pixels you need in order to draw a line
of a particular screen length):

Yes, you would need to know the resolution of your screen. Unfortunately for
you, it is difficult to impossible to correctly determine the resolution
programmatically. There is a handle graphics root property which will tell you
the screen resolution as known to the operating system, but what the operating
system knows and what the truth is can be completely different (and often are.)

Even if you had a copy of the specifications for your monitor, you would need
to know that it still met those specifications (probably not for an aging
CRT), and you would have to know that the display has been set exactly as
calibrated: especially on CRTs, it is not uncommon to find that the HORZ or
VERT size has been fiddled with by some user of the monitor. LCD and LED
screens should not suffer from any bowing, but CRTs certainly can. And if you
have a CRT in a decent magnetic field...



So, what you should do is create the largest figure you can, draw on it the
largest line you can of known pixel length, and then take a precision
measuring device and measure it as accurately as you can. And hope that the
next user along doesn't fiddle with the controls.
From: Samiov on 16 May 2010 03:00
ImageAnalyst <imageanalyst(a)mailinator.com> wrote in message <b80c94af-8344-4af4-850b-eaea76c19f36(a)h9g2000yqm.googlegroups.com>...
> No, the screen has nothing to do with it. Your image could take up
> the same, say, 1024 pixels across your screen but if it's an image of
> a cell in a microscope or a galaxy from a telescope they'll have
> different real world sizes. You need some calibration target in the
> plane of the scene. This can be the image or a ruler, or a micron bar
> that the microscope puts on there, or a known distance or size of some
> known object in your scene such as the angular subtense between two
> stars. What do you have in your scene that has a known size?
__________________________________________________________________________
But If I know the zoom of my image...I used a microscope caméra which enlarge my image 50 times...does it help to know the size of a pixel ? or not?

In my scene I have the woven fabrics that you know and the only thing I can extract is maybe the yarn diameter but it changes from an image to another
From: ImageAnalyst on 16 May 2010 09:11
On May 16, 3:00 am, "Samiov " <Samy...(a)yahoo.fr> wrote:
> But If I know the zoom of my image...I used a microscope caméra which enlarge my image 50 times...does it help to know the size of a pixel ? or not?
>
> In my scene I have the woven fabrics that you know and the only thing I can extract is maybe the yarn diameter but it changes from an image to another
---------------------------------------------------------------------------------
You still need to know the size of your scene corresponding to one
pixel in your digital image (not screen pixel). So you still need to
know the size of your total field of view (or some known distance
smaller than that) and the number of pixels across that distance in
your image. The 50X microscope mag doesn't help with that. If your
microscope system can stamp your image with a "micron bar" or whatever
they call it, that would be good. Otherwise you'll have to put in
something of a known scale, such as one of these perhaps:
http://www.edmundoptics.com/search/index.cfm?criteria=scales&x=0&y=0

Or you can use calibrated microspheres:
http://www.brumleysouth.com/calibration_standards.php
http://www.2spi.com/catalog/standards/microspheres.shtml
http://www.appliedphysicsusa.com/PSLSpheres-3000Series.html

 |  Next  |  Last
Pages: 1 2
Prev: Standalone Parallel Computing
Next: Problem with Move Camera Forward/Back using Camera Toolbar