10.126. Remove Minimum Size Features

Group (Subgroup)

Processing (Cleanup)

Description

This Filter removes small Features from a segmented dataset and fills in the space they leave behind by expanding their neighbors. Features whose cell count is below the user-specified Minimum Allowed Features Size are erased; the remaining features grow outward into the erased cells until every cell is reassigned to a surviving feature.

This is typically used immediately after segmentation (Segment Features (Misorientation) or similar) to discard spurious single-cell or few-cell “grains” that are almost always noise rather than real features.

How This Filter Works

  1. For each Feature, look up its cell count from the input Feature Num. Cells array.

  2. If the feature’s count is below the threshold, mark it for removal. All of its cells are temporarily set to Feature Id 0 (“unassigned”).

  3. After all small features are removed, the remaining features are isotropically coarsened to fill the gaps.

What is Isotropic Coarsening?

“Isotropic coarsening” is an iterative dilation that grows each surviving feature outward by one cell layer at a time, uniformly in all directions. On each pass, an unassigned cell is given the Feature Id of whichever neighbor feature touches it; if multiple features touch it, one is picked. The process repeats until no unassigned cells remain. The result is that every removed cell is absorbed into the closest surviving feature, with no preferred direction.

Minimum Size Units

The Minimum Allowed Features Size is in cells (integer voxel count), not physical units. A threshold of 10 means “remove any feature with fewer than 10 cells,” regardless of the image’s spacing. To convert a physical-volume threshold into a cell count, divide by the product of the cell spacing (dx * dy * dz).

Entering a number larger than the size of the largest feature produces an error, since every feature would be removed. If the size distribution is unknown, compute feature sizes first with Compute Feature Sizes and inspect the range.

Apply to Single Phase

When Apply to Single Phase is enabled, the size threshold is applied only to features of the specified phase. Features of other phases are untouched regardless of their size. This is useful when small features are noise in one phase but meaningful in another (e.g., small precipitates in a matrix).

WARNING: Feature Data Will Become Invalid

By modifying the cell-level Feature Ids, any feature-level data that was previously computed (sizes, centroids, average orientations, etc.) will almost certainly be invalid after this filter runs. Re-run any feature-level computation filters downstream of this one to ensure accurate results.

WARNING: NeighborList Removal

If the Cell Feature Attribute Matrix contains any NeighborList arrays, they will be removed because the list of neighbors for each surviving feature has changed. Re-run Compute Feature Neighbors afterward to rebuild the neighbor relationships.

Required Input Sources

Input Parameter(s)

Parameter Name

Parameter Type

Parameter Notes

Description

Minimum Allowed Features Size

Scalar Value

Int64

Minimum allowed features size

Apply to Single Phase

Bool

Apply to Single Phase

Phase Index

Scalar Value

Int32

Target phase to remove

Input Cell Data

Parameter Name

Parameter Type

Parameter Notes

Description

Input Image Geometry

Geometry Selection

Image

The input image geometry (cell)

Cell Feature Ids

Array Selection

Allowed Types: int32 Comp. Shape: 1

DataPath to FeatureIds DataArray

Input Feature Data

Parameter Name

Parameter Type

Parameter Notes

Description

Feature Num. Cells Array

Array Selection

Allowed Types: int32 Comp. Shape: 1

DataPath to NumCells DataArray

Feature Phases

Array Selection

Allowed Types: int32 Comp. Shape: 1

DataPath to Feature Phases DataArray

Example Pipelines

  • (02) Small IN100 Full Reconstruction

DREAM3D-NX Help

If you need help, need to file a bug report or want to request a new feature, please head over to the DREAM3DNX-Issues GitHub site where the community of DREAM3D-NX users can help answer your questions.