Login  Register

Re: what is the reason for the difference seen here

Posted by Joel Sheffield on Sep 15, 2005; 4:35pm
URL: http://imagej.273.s1.nabble.com/what-is-the-reason-for-the-difference-seen-here-tp3704870p3704874.html

Jeff, and others,

As your post suggests, I think that it is helpful to distinguish
between the actual values of the pixels and the display values.  
Remember that the monitor screen is actually able to display only a
limited range of values (256 comes to mind).  As a result, ImageJ
will set an output LUT (i.e. the range of display pixels) to display
the range of the data in 1/256 intensity steps.  Thus, even if you
use a 16-bit input file, your display will still be 8-bits.  This
will not affect your data, just how it appears.  You can manually set
the display to a given output scale with the b/c routine.  

However, once you click on "apply" the values are reassigned
(although this requires an 8-bit image) and your data is changed.  

> On Sep 14, 2005, at 6:58 PM, sean_incali wrote:
>
> >> ImageJ does automatic contrast/brightness adjustment for display
> >> under some circumstances.  If you examine the histogram for the
> >> image, you'll find that it's unaltered.
> >>
> > http://img369.imageshack.us/img369/3179/compare6cb.jpg
> >
> >
> > How do you use the histogram to check? Are you comparing the
> > intensity distributions? (I guess that's the only thing I can think
> > of using Histogram to check...)
>
> Here's what I did to verify what I thought I'd seen:
>
> I opened a RAW image file in ImageJ, and noted its range of values (in
> this case, 0-32767 or thereabouts).
>
> I used Process: Math: Divide to divide all pixel values by 16.  The
> resulting image was very dark, as I'd expect.  I saved this image in a
> new file, again in RAW format.
>
> I re-opened the original file and the divided-by-16 file.  Both images
> *appeared* essentially identical as displayed by ImageJ.  However,
> while the original image had a range of 0-32767, the new image had a
> range of 0-2047.
>
> I suspect that the image you're examining has no fully-saturated
> pixels, and that LSM browser is displaying it without adjustment,
> while ImageJ is automatically adjusting contrast and brightness to
> maximize visual contrast.
>
> If LSM browser lets you examine the histogram of your image, I'd
> suggest examining the histogram there and in ImageJ.  I'm guessing
> that you'll see the same values and the same distribution in both
> cases.
>
> Note that I'm not an LSM user, so I can't vouch for this from personal
> experience.
>
> > Also I have a question about opening a lsm file using LSM plugin vs
> > using an image in Tiff format exported from LSM broswer.
> >
> > If I open the "Tiff file" in ImageJ, is that same as the lsm file
> > opened using the plugin? (same, I mean in terms of intensity in two
> > channels.)
> >
> > If they are different, how different are they? in what respect?
>
> Again, I don't know what LSM does.  However, I do believe that ImageJ
> does the same autoscaling behavior with TIFFs that it does with RAW
> data. --
>  -jeffB (Jeff Brandenburg, Duke Center for In-Vivo Microscopy)


Joel B. Sheffield, Ph.D.
Biology Department, Temple University
1900 North 12th Street
Philadelphia, PA 19122
[hidden email]  
(215) 204 8839, fax (215) 204 0486
http://astro.temple.edu/~jbs