LabelOverlaps

LabelOverlaps counts pixel co-occurrences between two 2D label images. It produces one output row per (reference_label, spot_label) pair observed in the union of non-zero pixels.

Inputs are label_image and reference_image. Outputs are reference_label, spot_label, and overlap_count. Both images must have compatible shape and label semantics.

Use it to connect detected spots or predicted objects to reference objects. The tool does not solve matching by itself; it reports raw overlap counts that can be filtered or summarized downstream.

Dependencies and Core Libraries

BioImageFlow core APIs, imageio, and NumPy.

Assumptions

Both label images are aligned 2D arrays with compatible shape.

Minimal Example

from bioimageflow_core import Arguments
from bioimageflow_common_tools import LabelOverlaps

LabelOverlaps().process_row(
    Arguments(label_image="spots.tif", reference_image="nuclei.tif")
)

Expected Results

The output table contains one row per observed label pair and its pixel overlap count.

Failure Modes

Mismatched shapes can fail during NumPy indexing; missing files fail through imageio.