Re: Wand and XStart,YStart

Posted by Wayne Rasband on Jul 08, 2009; 2:34am
URL: http://imagej.273.s1.nabble.com/Wand-and-XStart-YStart-tp3691871p3691872.html

This bug, introduced in v1.43a, is fixed in the v1.43c daily build.  
The XStart and YStart values saved by the particle analyzer were not  
correct.

-wayne


On Jul 7, 2009, at 7:04 PM, Ben Tupper wrote:

> Hi All,
>
> I am still experiencing some bad magic using the new Wand -  
> specifically, I wonder if the inputs to the doWand(x,y) macro  
> command have changed with the version.  I am 99% sure that in the  
> past (at least in version 1.42) I provided doWand(x,y) with  
> unscaled pixel coordinates even if the image had a scale.  In  
> general I used pixel coordinates taken from the Results Table when  
> the XStart and YStart coordinates were available.  However, it  
> doesn't seem to me that this behavior is the same in version 1.43b  
> even if I am using the wand in legacy mode.
>
> Below is a short macro that loads a small scaled binary image from  
> my web space (yeah for run("URL..."))  The macro loads for me with  
> the following scale info...
>
>
> Title: calROI-3.tif
> Width:  75.06 µm (88)
> Height:  65.68 µm (77)
> Resolution:  1.172 pixels per µm
> ID: -17
> Coordinate origin:  0,0
> Bits per pixel: 8 (inverting grayscale LUT)
> Display range: 0-255
> No Threshold
> Uncalibrated
> Path: calROI.tif
>
> - Running in version 1.43b, you can see that the entire image is  
> selected which is not what I expected.
>
> - Running in version 1.42q, you can see that the particle in the  
> upper left is highlighted which is what I expected.
>
> Have I missed something in the change of Wand?  My first guess is  
> that Wand is getting coords that have been changed into "real"  
> world units, but a quick peek at Macro.Functions.doWand() and  
> IJ.doWand() don't reveal anything funny to my eyes.  My second  
> guess is that the [XStart, YStart] pixel values [16,4] are  
> converted to a pixel just outside the masked pixels as [13.652  
> 3.413].  Zooming into that location shows it is indeed a background  
> pixel.
>
> If it is the latter, then it seems I can't use XStart, YStart with  
> Wand.  Ouch.
>
> Perhaps it is because XStart, YStart appear to be on the right hand  
> side of the object and Wand looks at the seed location and to its  
> right for an initial boundary pixel?
>
>
> // START HERE
> if (startsWith(getVersion(), "1.43")){
> run("Wand Tool...", "mode=Legacy tolerance=0");
> }
> run("URL...", "url=http://www.tidewater.net/~pemaquid/calROI.tif");
> run("Set Measurements...", "area bounding shape redirect=None  
> decimal=3");
> run("Analyze Particles...", "size=0-Infinity circularity=0.00-1.00  
> show=Nothing display clear record");
> x = getResult("XStart", 0);
> y = getResult("YStart", 0);
> print("Start [x,y]=", x, ", ", y);
> doWand(x,y);
> // END HERE
>
> Thanks and cheers,
> Ben
>
> Ben Tupper
>
>
> P.S. I sure hope you all are getting the sun that we here in Maine  
> are not getting this summer. In fact, we are huddled around the  
> fireplace today as it only hit 53F.  Brrr.  The only good thing  
> about it is that I don't have to paint the house!