| Average outline of several contours [message #37021] |
Thu, 20 November 2003 21:22  |
Beat.Schmutz
Messages: 6
Registered: October 2003
|
Junior Member |
|
|
I have combined 8 contours of bone cross-sections into one image
(individual ROIs in a IDLgrROIGroup). Now I like to determine the
path/outline which represents the average of these 8 contours. Is
anyone aware of a program/code/method that enables me to achieve this?
Any help would be greatly appreciated.
Beat Schmutz
|
|
|
|
| Re: Average outline of several contours [message #37073 is a reply to message #37021] |
Mon, 24 November 2003 19:44  |
Beat.Schmutz
Messages: 6
Registered: October 2003
|
Junior Member |
|
|
> Personally, I think averaging in contour space is a problem. Contours
> are sensitive to any little fluctuation in the underlying image. For
> example, a noise fluctuation in a single pixel might produce small
> extra contour loops.
>
> Instead, I would say you should add the underlying images together,
> and then find the master contour from the mean image.
>
> Good luck!
> Craig
Thanks Craig.
Beat
|
|
|
|
| Re: Average outline of several contours [message #37103 is a reply to message #37021] |
Sun, 23 November 2003 12:19 |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
Beat.Schmutz@flinders.edu.au (Beat Schmutz) writes:
>> Assuming the contours are roughly circular, you could do this:
>
> The contours are roughly circular for the shaft region of the bone but
> not for the ends (condyles). Sorry, I should have mentioned that in
> my original posting.
Personally, I think averaging in contour space is a problem. Contours
are sensitive to any little fluctuation in the underlying image. For
example, a noise fluctuation in a single pixel might produce small
extra contour loops.
Instead, I would say you should add the underlying images together,
and then find the master contour from the mean image.
Good luck!
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@REMOVEcow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
| Re: Average outline of several contours [message #37106 is a reply to message #37021] |
Sun, 23 November 2003 03:42 |
Beat.Schmutz
Messages: 6
Registered: October 2003
|
Junior Member |
|
|
> Assuming the contours are roughly circular, you could do this:
The contours are roughly circular for the shaft region of the bone but
not for the ends (condyles). Sorry, I should have mentioned that in
my original posting.
>
> Average the radial values of the contours at each regular theta.
Thanks for your suggestion, I'll keep it in mind.
Cheers,
Beat
|
|
|
|
| Re: Average outline of several contours [message #37114 is a reply to message #37021] |
Fri, 21 November 2003 07:14 |
K. Bowman
Messages: 330
Registered: May 2000
|
Senior Member |
|
|
In article <f40e34db.0311202122.37499b6@posting.google.com>,
Beat.Schmutz@flinders.edu.au (Beat Schmutz) wrote:
> I have combined 8 contours of bone cross-sections into one image
> (individual ROIs in a IDLgrROIGroup). Now I like to determine the
> path/outline which represents the average of these 8 contours. Is
> anyone aware of a program/code/method that enables me to achieve this?
> Any help would be greatly appreciated.
>
> Beat Schmutz
Assuming the contours are roughly circular, you could do this:
Compute the centroid (average position) of all the points in all the
contours.
Transform the contours to polar coordinates centered on the centroid.
Interpolate each contour to regularly spaced points in theta.
Average the radial values of the contours at each regular theta.
Transform back to Cartesion coordinates, if necessary.
The only tricky thing I see is doing the interpolation in theta across
the "seam" (where theta equals both 0 and 2 pi). That should be easy to
fix by rotating the coordinates by pi for points near the seam.
Ken Bowman
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
