FilterLabels¶

FilterLabels removes labels that fail area, border-contact, intensity, or shape constraints, then relabels retained objects sequentially.

Inputs include labels, min_area, max_area, remove_border_touching, optional intensity_image, min_mean_intensity, min_solidity, and max_eccentricity. Outputs are output_labels and object_count.

Minimal Example¶

from bioimageflow_core import Arguments
from bioimageflow_segmentation_tools import FilterLabels

result = FilterLabels().process_row(
    Arguments(labels="labels.tif", output_labels="filtered.tif", min_area=20)
)

Inputs¶

labels: 2D or 3D label image.
min_area and max_area: pixel/voxel count limits.
remove_border_touching: remove objects touching an image border.
intensity_image and min_mean_intensity: optional intensity filter.
min_solidity and max_eccentricity: 2D shape filters.

Outputs¶

output_labels: filtered labels relabeled sequentially.
object_count: number of retained objects.

Dependencies and Core Libraries¶

imageio, NumPy, and scikit-image region-properties measurement.

Assumptions¶

Background is 0. Solidity and eccentricity are applied only to 2D inputs; 3D inputs still support area, border, and intensity filters.

Expected Results¶

Synthetic fixtures remove small, border-touching, or low-intensity objects and return sequential label IDs.

Failure Modes¶

Shape mismatches between labels and intensity images, unreadable files, and write failures raise errors.