Dear imagej-users
I have two problems about size filter. The content is a little long, but Please!!! Please look at it patiently. This problem has puzzled me for a long time…
1) Can someone help me to check my size filter workflow is right or not?
I want to give my segmentation result clean up.
my sample’s foreground is grains(i.e. soil), and the background is voids(i.e. pores).
I would like to remove the all isolated cluster grains in the background, so I used the Process- Binary-Fill hole to achieve this.
Then I invert the images to make the foreground is voids(because this is my final target to analysis), and the background is grains. Now I want to remove the void isolated cluster whose size is smaller than 10 voxels. I use the Plugin-BoneJ-Analyse Particles to set the size filter is 10 to do this.
This is my simple macro for your reference:
run(“Fill Holes”, “stack”);
run(“Invert”, “stack”);
run(“Particle Analyser”, “surface_area enclosed_volume moments euler thickness mask ellipsoids min=10.000 max=Infinity surface_resampling=2 show_particle show_surfaces show_centroids show_axes show_ellipsoids show_stack surface=Gradient split=0.000 volume_resampling=2”);
//setTool(“hand”);
I am a beginner in ImageJ, and I am afraid of some logical problems, so I would like to ask you to help me to have a look that the above idea is right? Or, do you have any better suggestions?
2) If the above operation is correct, then I will get the final size filter result from Plugin-BoneJ-Analyse Particles. But how can I transform this result images into an 8-bit images for the later analysis?
I tried Image > Adjust > Threshold to binarize the particle label image named like “XX_parts”, but the result is not good. Because some details will be lost and cannot be binarized, that is to say, there is a loss of objects between binarized images and the original “XX_parts” images.
And I find it strange in Analyse Particles results that there are also differences between “XX_parts” image and “XX_volume” image. To be honest, I don't quite understand. I think both of them should be the same, just that one shows the label ID information and the other shows the volume information. So the idea that I want to use volume image to do binarization seems to be infeasible.
Does anyone know how to solve this?
Any kind of help will be greatly appreciated!
Mengying
| |
Mengying Ni
|
|
[hidden email]
|
签名由网易邮箱大师定制
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
