| Moving an ROI [message #28325] |
Tue, 04 December 2001 13:35  |
Andre Kyme
Messages: 19
Registered: September 2001
|
Junior Member |
|
|
Hi everybody,
Anyone know of a routine to move an ROI around in a draw widget? I've
collected the ROI coordinates and now want to adjust its position. Wrote
a bit of code that's kind of working but is very dodgy when the ROI gets
to edges. Thanks for the help,
Andre Kyme
Dept Medical Physics
Westmead Hospital
|
|
|
|
|
|
| Re: Moving an ROI [message #28403 is a reply to message #28325] |
Wed, 05 December 2001 13:55  |
Andre Kyme
Messages: 19
Registered: September 2001
|
Junior Member |
|
|
Thanks Pavel,
I had rewritten the move ROI program yesterday and it seems to be more robust. I
use the set_graphics_function argument (to the device), and set this equal to 6
(XOR mode) so that the ROI "floats" over the image. When a new position is
detected, the old position is replotted to make it disappear and the new position
plotted. This is all OK.
Now about the edges:
As you say, by defining a new ROI equal to (Xverts > 0)AND(Xverts < xMax), and
similarly for Y, we create a new ROI that fits in the draw widget and has its edge
along the draw widget boundary if there is any part of the ROI outside the allowed
region.
Another way would be to form a bit mask for the Xverts, setting equal to 1 wherever
the vertices are in the allowed region, and doing the same for the Yverts. If we
then AND these two masks and finally do a WHERE on the result, we can plot
Xverts[result], Yverts[result]. The problem I see with this is that if we are over
the boundary, the ROI stops at the point(s) of contact. I eventually want the total
counts inside the ROI, which means that I need to CLOSE it off first, before doing
a polyfillv. If I used this last method as is, the ROI would be truncated by simply
drawing a line from one point of contact to the other - what I really want is the
boundary to follow the edge of the draw widget where it has crossed the draw
boundary. Closing off the region isn't so simple as far as I can see. I think you
probably know what I'm talking about even if this explanation is incomprehensible!
Maybe I'll go for the first method.
Thanks for your reply Pavel,
Andre
|
|
|
|
| Re: Moving an ROI [message #28408 is a reply to message #28325] |
Wed, 05 December 2001 08:55 |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
Nobody replied. I guess a routine is not there. But it seems simple to
write one. Once you have a ROI, all it is is just a polygon. With motion
events on, just add delta X and Y to its Data property and redraw the view.
Edges need a special thought, seems to me. First, have you decided how
do *you* want them treated? If you trim the ROI, you need to redefine
Data because IDLgrPolygon has no *Range properties. I did this by
keeping the original ROI data in Uvalue, then trimming Data as Data =
[[X_min < Data[0,*] < X_max], [Y_min < Data[1,*] < Y_max], [Data[2,*]],
then redrawing.
If you want to expand the plot range as ROI gets dragged past the edge,
you need to rescale the plot and change axes limits. Although I
anticipate somewhat jerky screen updates with large plots, it also seems simple.
For data processing, just reuse the data from Uvalue (the original ROI).
Pavel
Andre Kyme wrote:
>
> Hi everybody,
>
> Anyone know of a routine to move an ROI around in a draw widget? I've
> collected the ROI coordinates and now want to adjust its position. Wrote
> a bit of code that's kind of working but is very dodgy when the ROI gets
> to edges. Thanks for the help,
>
> Andre Kyme
>
> Dept Medical Physics
> Westmead Hospital
|
|
|
|
| Re: Moving an ROI [message #28517 is a reply to message #28378] |
Thu, 06 December 2001 17:00 |
Andre Kyme
Messages: 19
Registered: September 2001
|
Junior Member |
|
|
Hi Pavel,
Not using objects I'm afraid, just run-of-the-mill stuff.
I've gone with the mask method and got it up and running - the trunction of the ROI at
the edges
occurs by drawing a line from the points of contact rather than following the boundary,
but this is
OK because the ROI is actually considered unacceptable unless the entire array of X and
Y vertices are within
the draw widget area.
Think it works OK - but I've said those famours last words before.
Maybe I'll look into objects next year,
Thanks,
Andre
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
