Posted by
Joachim Wesner on
Apr 03, 2006; 10:43am
URL: http://imagej.273.s1.nabble.com/ImageJ-and-Metallography-tp3703221p3703225.html
Gabriel Landini
<G.Landini@BHAM. An:
[hidden email]
AC.UK> Kopie:
Gesendet von: Thema: Re: Bug with complex ROIs
ImageJ Interest
Group
<
[hidden email]
.GOV>
03.04.2006 10:59
Bitte antworten
an ImageJ
Interest Group
Hi,
yep, but I thought this really is a bug and not a "feature" as it first
seemed to happen if the "relative coordinates" of the enclosing rect are
negative or extend over the image. It also does not only mean "recentering"
but might be also rescaling of the ROI shape, I do not yet understand what
exactly happens, especially what happens with polygon ROIs etc.??? Is this
behaviour documented somewhere?
The whole discussion might appear a bit "academic", however it happened to
be that I tested my plugin with images that show a circular pupil which in
some case just slightly extends over the corner, so I thought I could just
get the right ROI using an oversize circle, but that is just the condition
where the bug seems to show up.
On Monday 03 April 2006 09:47, Joachim Wesner wrote:
> When doing "Edit/Clear" as also "Edit/Clear outside" the selection
> corresponds to the ROI (i.e. circle with cut off sides), however, when
> doing "Image/Crop", the origin of the selection is "recentered" somehow,
> even up to the point (if possible) that a new fully circular ROI results,
> that, however, will no longer agree with the selection in the source
> window.
Also when changing the image type (8 bit to rgb, etc) the selection gets
centred too.
And if one tries to convert an 8 bit grey into 8 bit colour, the expected
converted error table comes up, but the selection is lost.
Cheers,
Gabriel
______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit
http://www.messagelabs.com/email
______________________________________________________________________