Image Cleaning¶

Image Cleaning removes features of a segmented image that can't take part in transport: pores that don't connect across the sample, and specks of solid floating in the pore space. Use it after segmenting, to analyse the connected pore space rather than every pore.

Find it Tools › Preparation › Image Cleaning, or Prepare on an image's page
Needs a segmented 3D image, or a batch of them
Returns the cleaned image, saved over the original or as a new image
Runs on CPU
Science Image Cleaning

Run it¶

Open Tools, then Preparation, and select Image Cleaning.
Under Source, choose the Image, or a batch to clean every image in it.
Under Settings, tick Fill Invalid Pores, Trim Floating Solid, or both. They run in that order.
Under Output, choose Overwrite Original or Save as New, and for a new image, its New image name. The panel below lists the images the job will create or overwrite.
Select Run. The job page opens; a batch makes one job per image and takes you to Jobs.

The Image Cleaning form with both operations ticked and a new image name ready. — Both operations ticked, saved as a new image named in the panel at the bottom.

Settings¶

Setting	Default	Range	What it does
Image		your team's images and batches	The image to clean, or a batch to clean one by one.
Fill Invalid Pores	off	on, off	Turns into solid every pore cluster that doesn't join two opposite faces of the image: closed pores and dead-end clusters.
Trim Floating Solid	off	on, off	Turns into pore every solid cluster that touches no face of the image.
Save mode	Overwrite Original	Overwrite Original, Save as New	Whether to replace the image or keep it and add a cleaned copy.
New image name	the image's name with `_cleaned`	up to 200 characters	Save as New only.
Backend			The software that cleans the image, PoreSpy.

Tick at least one operation.

Results¶

The job page shows an Output image panel, or Image updated when you overwrote the original, with View output image. The cleaned image's porosity is measured again, so compare it with the original's to see how much pore space was disconnected.

A small example: each option on a 100³ synthetic sample, shown on the middle slice. Left: a sample with little pore space, porosity 0.15; every pore cluster that doesn't span the image is filled. Right: a sample with little solid, porosity 0.85; every solid cluster that touches no face becomes pore. Clusters are found in 3D, so whether one is kept can depend on connections outside the slice.

A completed Image Cleaning job linking to the cleaned image. — A cleaning job that updated its image in place, and View output image.

Common problems¶

Saving over the image removed its 3D fields. Overwriting an image discards the fields its earlier results saved, such as velocity or concentration, though their numbers stay. Use Save as New to keep them.
Almost all the pore space disappeared. No pore cluster joins two opposite faces, so Fill Invalid Pores filled them all. Check the segmentation, or run a Connectivity Analysis on the image.
Porosity changed more than expected. Both operations change porosity on purpose: cleaning moves an image from total towards connected porosity.

Science¶

How each operation decides which clusters to keep, and what cleaning does to porosity and transport: Image Cleaning in the Scientific reference.