discrepancy between APi and IJ.run in Find maxima

[Aryeh Weiss]

We are running FInd Maxima with an ROI selected. Find Maxima "respects"
the ROI,
and only marks the maxima that  are within the ROI. I should note that
the ROI is an "irregular" shape
(ie, not rectangular). So the following code in python:

IJ.run("Find Maxima...", "noise=" + str(noiseThreshold)+" output=[Maxima
Within Tolerance]")

works as expected.

However, the following API version:

max_proc = MaximumFinder().findMaxima(ip, float(noiseThreshold),
ImageProcessor.NO_THRESHOLD, int(1), True, False)

marks an extra maximum which is outside of the ROI (but inside the
bounding rectangle).
Here is the link to the API:
http://imagej.nih.gov/ij/developer/api/ij/plugin/filter/MaximumFinder.html

It appears that the API version is using the bounding box of the ROI.
Is this a bug, or did I use it incorrectly?

--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: [phone]

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Rasband, Wayne (NIH/NIMH) [E]]

On Jul 16, 2014, at 8:01 AM, Aryeh Weiss wrote:

> We are running FInd Maxima with an ROI selected. Find Maxima "respects" the ROI,
> and only marks the maxima that  are within the ROI. I should note that the ROI is an "irregular" shape
> (ie, not rectangular). So the following code in python:
>
> IJ.run("Find Maxima...", "noise=" + str(noiseThreshold)+" output=[Maxima Within Tolerance]")
>
> works as expected.
>
> However, the following API version:
>
> max_proc = MaximumFinder().findMaxima(ip, float(noiseThreshold), ImageProcessor.NO_THRESHOLD, int(1), True, False)
>
> marks an extra maximum which is outside of the ROI (but inside the bounding rectangle).
> Here is the link to the API:
> http://imagej.nih.gov/ij/developer/api/ij/plugin/filter/MaximumFinder.html
>
> It appears that the API version is using the bounding box of the ROI.
> Is this a bug, or did I use it incorrectly?

I am unable to duplicate this problem. The following macro and JavaScript produce exactly the same output. When reporting a potential bug, it is very helpful to include a small, runnable example macro or script that reproduces the problem.

  // Macro
  run("Cell Colony (31K)");
  run("Invert");
  makeOval(133, 85, 110, 243);
  run("Find Maxima...", "noise=50 output=[Maxima Within Tolerance] exclude");

  // JavaScript
  imp = IJ.openImage("http://imagej.nih.gov/ij/images/Cell_Colony.jpg");
  ip = imp.getProcessor();
  ip.invert();
  roi = new OvalRoi(133, 85, 110, 243);
  ip.setRoi(roi);
  ip2 = (new MaximumFinder()).findMaxima(ip, 50, 1, true)
  new ImagePlus("ip2", ip2).show();

-wayne

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Aryeh Weiss]

Wayne replied to me offlist, but my copy to the list bounced because of
the attachment. For the benefit of the list, I will summarize his reply.

On 7/17/14, 12:10 AM, Rasband, Wayne (NIH/NIMH) [E] wrote:

> On Jul 16, 2014, at 8:01 AM, Aryeh Weiss wrote:
>
>> We are running Find Maxima with an ROI selected. Find Maxima "respects" the ROI,
>> and only marks the maxima that  are within the ROI. I should note that the ROI is an "irregular" shape
>> (ie, not rectangular). So the following code in python:
>>
>> IJ.run("Find Maxima...", "noise=" + str(noiseThreshold)+" output=[Maxima Within Tolerance]")
>>
>> works as expected.
>>
>> However, the following API version:
>>
>> max_proc = MaximumFinder().findMaxima(ip, float(noiseThreshold), ImageProcessor.NO_THRESHOLD, int(1), True, False)
>>
>> marks an extra maximum which is outside of the ROI (but inside the bounding rectangle).
>> Here is the link to the API:
>> http://imagej.nih.gov/ij/developer/api/ij/plugin/filter/MaximumFinder.html
>>
>> It appears that the API version is using the bounding box of the ROI.
>> Is this a bug, or did I use it incorrectly?
> I am unable to duplicate this problem. The following macro and JavaScript produce exactly the same output. When reporting a potential bug, it is very helpful to include a small, runnable example macro or script that reproduces the problem.
>
>    // Macro
>    run("Cell Colony (31K)");
>    run("Invert");
>    makeOval(133, 85, 110, 243);
>    run("Find Maxima...", "noise=50 output=[Maxima Within Tolerance] exclude");
>
>    // JavaScript
>    imp = IJ.openImage("http://imagej.nih.gov/ij/images/Cell_Colony.jpg");
>    ip = imp.getProcessor();
>    ip.invert();
>    roi = new OvalRoi(133, 85, 110, 243);
>    ip.setRoi(roi);
>    ip2 = (new MaximumFinder()).findMaxima(ip, 50, 1, true)
>    new ImagePlus("ip2", ip2).show();
>
> -wayne
>
>

The relevant part of  my script was:

ip = imp.getProcessor()
max_proc = MaximumFinder().findMaxima(ip, float(noiseThreshold),
ImageProcessor.NO_THRESHOLD, int(1), True, False)

Where the image had an ROI selected.

The mistake is that I did nto transfer the nonrectangular ROI to the
image processor.

ip = imp.getProcessor())
ip.setRoi(imp.getRoi())
max_proc = MaximumFinder().findMaxima(ip, float(noiseThreshold),
ImageProcessor.NO_THRESHOLD, int(1), True, False)

solves the problem.

--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: [phone]

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Michael Schmid]

Hi Aryeh,

in a PlugInFilter, you the ROI of the ImageProcessor is always equal to that of the ImagePlus.  If you write a PlugIn, you can't be sure about this; you have to care yourself about setting the Roi for the ImageProcessor in case you need it:

For area rois, this will typically look like the following:

  Roi roi = imp.getRoi();
  if (roi!=null && roi.isArea())
     ip.setRoi(roi);
  else
     ip.setRoi((Roi)null);


Michael
________________________________________________________________
On Jul 16, 2014, at 14:01, Aryeh Weiss wrote:

> We are running FInd Maxima with an ROI selected. Find Maxima "respects" the ROI,
> and only marks the maxima that  are within the ROI. I should note that the ROI is an "irregular" shape
> (ie, not rectangular). So the following code in python:
>
> IJ.run("Find Maxima...", "noise=" + str(noiseThreshold)+" output=[Maxima Within Tolerance]")
>
> works as expected.
>
> However, the following API version:
>
> max_proc = MaximumFinder().findMaxima(ip, float(noiseThreshold), ImageProcessor.NO_THRESHOLD, int(1), True, False)
>
> marks an extra maximum which is outside of the ROI (but inside the bounding rectangle).
> Here is the link to the API:
> http://imagej.nih.gov/ij/developer/api/ij/plugin/filter/MaximumFinder.html
>
> It appears that the API version is using the bounding box of the ROI.
> Is this a bug, or did I use it incorrectly?
>
> --aryeh
>
> --
> Aryeh Weiss
> Faculty of Engineering
> Bar Ilan University
> Ramat Gan 52900 Israel
>
> Ph:  972-3-5317638
> FAX: [phone]
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html