Prev: roc in matlab
Next: Matrix power symbolic
From: Ankit Goyal on 3 Jun 2010 09:50
I have a image which is needed to get thresholded for further processing.Please help me with that!!
Thanks in advance

http://drop.io/imagest
From: Walter Roberson on 3 Jun 2010 10:08
Ankit Goyal wrote:
> I have a image which is needed to get thresholded for further
> processing.Please help me with that!!

> http://drop.io/imagest

If you have the image processing toolbox, greythresh()
From: Ankit Goyal on 3 Jun 2010 10:15
@Walter Roberson

ALready tried graythresh as well as local adaptive thresholding but as u can see the background pixels and foregrnd pixels are similar in intensities so graythresh will not work here ..I want smthg robust that can threshold the cell boundaries..
Any other suggestion!!
From: Walter Roberson on 3 Jun 2010 10:24
Ankit Goyal wrote:
> @Walter Roberson
>
> ALready tried graythresh as well as local adaptive thresholding but as
> u can see the background pixels and foregrnd pixels are similar in
> intensities so graythresh will not work here ..I want smthg robust that
> can threshold the cell boundaries..
> Any other suggestion!!

There is no such thing as robust unsupervised thresholding, as
appropriate thresholds are a matter of context rather than of absolute
value.

Take an astronomical photograph and automatically threshold it with
intent to study the stars; now take the same astronomical photograph and
automatically threshold it with intent to study the cosmic background
radiation. What was noise to be discarded for the first usage is now
information for the second usage -- so clearly there cannot be any
automated method that, without input from you as to what is foreground
and what is background, decides correctly in both cases.

Are you attempting to greythresh the image that has all those white
lines on it? If so I am not surprised that it treats the entire rest of
the image as background. Have you tried greythresh on the original image
before it had the white lines added?
From: Ankit Goyal on 3 Jun 2010 10:58
@Walter Roberson
Actually the curvy white lines are the one which r needed for the analysis and the rest u can take as background.And i know that image processing is based on perception ,just saying that the image should be well thresholded as to further analysis.
 |  Next  |  Last
Pages: 1 2
Prev: roc in matlab
Next: Matrix power symbolic