[ImageJ-bugs] [Bug 988] New: Error in perimeter evaluation for particle analyzer

bugzilla at fiji.sc bugzilla at fiji.sc
Wed Jan 14 11:06:03 CST 2015


http://fiji.sc/bugzilla/show_bug.cgi?id=988

            Bug ID: 988
           Summary: Error in perimeter evaluation for particle analyzer
           Product: Fiji
           Version: unspecified
          Hardware: PC
                OS: Windows
            Status: NEW
          Severity: normal
          Priority: P5
         Component: ImageJ2
          Assignee: imagej-bugs at imagej.net
          Reporter: pippomaggi at gmail.com

Created attachment 228
  --> http://fiji.sc/bugzilla/attachment.cgi?id=228&action=edit
Squares with the same area but different rotations

The computation of the perimeter for particle analyzer is not consistent when
you analyze particles with edges that are not parallel, perpendicular, or do
not have a 45 degree inclination.

The issue can be reproduced as follows:

1 - With an external program, generate an 8-bit image with a series of squares
with the same area but different rotation from 0 to 45° (example is provided)

2 - Threshold the image (Image->Adjust->Threshold)

3 - Apply particle analysis to the image (Analyze->Analyze Particles)

The results will show same area but different perimeter.
The problem is inherited by shape descriptors based on Perimeter quantity and
plugins (e.g. Shape Descriptor1u).

The issue seems to be related to on edge detection because the outlined profile
is stepped and not continuous. Increase of image resolution does not improve.

Fiji ImageJ 2.0.0-rc-15/1.49k
Java 1.7.0_67
Windows 7 (64 bit)

The bug was found on ImageJ 1.44 as well.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://imagej.net/pipermail/imagej-bugs/attachments/20150114/15e6b3b8/attachment.html>


More information about the Imagej-bugs mailing list