Home »
Public Forums »
archive »
Image Segmentation
| Image Segmentation [message #41821] |
Wed, 24 November 2004 13:52 |
praveshsubramanian
Messages: 15
Registered: September 2004
|
Junior Member |
|
|
Hello All:
The edge detection operators, robert and sobel dont seem to work too
fine for me. I am working with MRI images of human body and the
problem is that the regions ( and of course, their boundaries) in the
images dont really stand out. As a result, if i use region-grow,
inevitably there is a leakage into some other region.
this problem is inherent as the human tissues are really not
distributed in an orderly manner. Filtering : tried all of them! they
all modify properties of the image (such as the number of pixels
belonging to intensity ranges).
the idea is to work on the raw image and get data out of it (smoothing
et al simply change the image properties). but it seems that we need
proper edge detection.
i think region_grow routine using thresholding works, but not for all
images for the same reason cited above. standard dev. multiplier is
way too aberrant, sometimes it simply covers the entire image instead
of a region.
What options does that leave me for? hm... either i write my own
routine using hints from u all.. or figure out a way using IDL
routines..
the idea is simply click on one region and only pixels from that
region get highlighted. the standard deviation of the pixel
intensities is around 80.
(magic!!!)
does someone have a magic wand???
Regards
PRavesh
|
|
|
|
Current Time: Fri Oct 10 08:34:53 PDT 2025
Total time taken to generate the page: 0.48265 seconds
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
