|
Dear All,
I would like to announce an updated version of the GrayMorphology plugin. It can be found at: http://rsb.info.nih.gov/ij/plugins/gray-morphology.html There are signigicant improvements in the performance of the morphologic operations. New features: -- "fast" versions of the erosion, dilation, opening and closing operations based on local histograms. Also the van Herk/Gil-Werman (vHGW) algorithm for linear SEs was implemented (it is really fast). -- Processing of stacks is now enabled -- The plugin can handle arbitrary ROIs best regards Dimiter |
|
On Tuesday 06 June 2006 18:07, Dimiter Prodanov wrote:
> I would like to announce an updated version of the GrayMorphology plugin. Thanks, very useful. How does one create a free form SE? If I choose free form I get java.lang.NullPointerException at mmorpho.StructureElement.calcVect(StructureElement.java:467) at mmorpho.StructureElement.<init>(StructureElement.java:121) at GrayMorphology_.run(GrayMorphology_.java:212) at ij.IJ.runFilterPlugIn(IJ.java:196) at ij.IJ.runUserPlugIn(IJ.java:264) at ij.IJ.runPlugIn(IJ.java:116) at ij.Executer.runCommand(Executer.java:95) at ij.Executer.run(Executer.java:46) at java.lang.Thread.run(Thread.java:619) Also I get a bit slower processing with the "fast erode" than with "erode". Is that OK or just the name of the algorithm has been swapped? t=getTime(); run("Gray Morphology", "radius=10 type=circle operator=erode"); print (getTime-t); run("Undo"); t=getTime(); run("Gray Morphology", "radius=10 type=circle operator=[fast erode]"); print (getTime-t); run("Undo"); t=getTime(); run("Minimum...", "radius=10"); print (getTime-t); run("Undo"); Runing the above macro with the 512x512 lena image produces: 1885 2657 2005 Any ideas? Cheers, Gabriel |
|
In reply to this post by Dimiter Prodanov
Dear Gabriel,
Fro the free-form SE - this is a bug. I'll look into it. In principle it is space delimited each line ends with the system newline character: \r\n - MS Win or \n - UNIX like OS As for the performance, the fast operations are fast only with large SEs: >See what the macro does with my favourite blobs image (256 x 254): > >Benchmarks Gray Morphology plugin >Gray Morphology erode r=5 t=991 >Gray Morphology fast erode r=5 t=3365 >Minimum filter r=5 t=421 > >Gray Morphology erode r=10 t=3615 >Gray Morphology fast erode r=10 t=4286 >Minimum filter r=10 t=3625 > >Gray Morphology erode r=15 t=7691 >Gray Morphology fast erode r=15 t=4877 >Minimum filter r=15 t=6169 > >Gray Morphology erode r=20 t=8643 >Gray Morphology fast erode r=20 t=2994 >Minimum filter r=20 t=7491 > >Gray Morphology erode r=25 t=8993 >Gray Morphology fast erode r=25 t=3365 >Minimum filter r=25 t=13539 > >Gray Morphology erode r=30 t=25427 >Gray Morphology fast erode r=30 t=3695 >Minimum filter r=30 t=17715 > >Cheers > >Dimiter |
|
Hi Dimiter,
> Fro the free-form SE - this is a bug. I'll look into it. In principle > it is space delimited > each line ends with the system newline character: \r\n - MS Win or > \n - UNIX like OS Ok, when I choose free form, there is no further dialog opening for inserting/typing the kernel (I suppose that this is what it does?) > As for the performance, the fast operations are fast only with large SEs: I see, thanks for the clarification. This may be useful to add to the plugin description. Cheers, Gabriel |
| Free forum by Nabble | Edit this page |
