LocalThresholdSegment

LocalThresholdSegment computes a Sauvola adaptive threshold with skimage.filters.threshold_sauvola, applies an optional offset, and labels connected foreground components.

Inputs are input_image, block_size, k, offset, and above. block_size must be an odd integer greater than or equal to 3.

Minimal Example

from bioimageflow_core import Arguments
from bioimageflow_segmentation_tools import LocalThresholdSegment

result = LocalThresholdSegment().process_row(
    Arguments(input_image="image.tif", labels="labels.tif", block_size=15, k=0.2)
)

Inputs

• input_image: 2D intensity image.

• block_size, k, and offset: local threshold parameters.

• above: when true, pixels above the local threshold are foreground.

Outputs

• labels: connected-component label image.

• object_count: number of foreground components.

Dependencies and Core Libraries

imageio, NumPy, and scikit-image Sauvola thresholding/labeling functions.

Assumptions

The image is 2D and local contrast separates foreground from background.

Expected Results

Synthetic fixtures with local bright objects produce stable component counts under the documented parameters.

Failure Modes

Invalid block sizes, unreadable images, unsupported dimensions, and write failures raise errors.