PostprocessLabels¶
PostprocessLabels removes labels smaller than min_size and relabels the
remaining objects sequentially from 1.
Inputs are labels and min_size. Outputs are output_labels and
object_count. The tool preserves image shape and writes a UInt32-style label
array through imageio.
Use it after any segmentation tool when small artifacts should be removed or
label IDs need to be compact. Label 0 is always background.
Dependencies and Core Libraries¶
BioImageFlow core APIs, imageio, and NumPy.
Assumptions¶
Input labels are integer-like arrays, label 0 is background, and size is
measured in pixels or voxels.
Minimal Example¶
from bioimageflow_core import Arguments
from bioimageflow_segmentation_tools import PostprocessLabels
PostprocessLabels().process_row(Arguments(labels="labels.tif", min_size=16))
Expected Results¶
Objects below min_size are removed and remaining object IDs are relabeled
sequentially.
Failure Modes¶
Missing or unsupported label images fail through imageio. Very high min_size
can remove all labels.