MedianDenoising

Purpose

MedianDenoising applies SAIRPICO median filtering through simgmedian2d, simgmedian3d, or simgmedian4d. It is a simple restoration and preprocessing tool for removing sparse impulse-like noise before segmentation, spot detection, or deconvolution.

Inputs

• input_image: intensity image in png, tif, or tiff format.

• denoising_type: 2D, 3D, or 4D. Default 2D.

• radius_x: median-filter radius in X. Default 2.

• radius_y: median-filter radius in Y. Default 2.

• radius_z: median-filter radius in Z for 3D and 4D modes. Default 1.

• radius_t: median-filter radius in time for 4D mode. Default 1.

• padding: whether to process border pixels using padding. Default False.

Outputs

• output_image: median-filtered intensity image. The default template is {input_image.stem}_filtered{ext}.

Assumptions

• The selected dimensional mode matches the input image layout.

• Radius values are tuned to preserve objects of interest.

• radius_z is ignored in 2D mode and radius_t is used only in 4D mode.

Dependencies and Core Libraries

• BioImageFlow core tool abstractions.

• SAIRPICO simglib environment with bioimageit::simglib==0.1.2.

• External commands: simgmedian2d, simgmedian3d, and simgmedian4d.

Minimal Example

from bioimageflow_core import Arguments
from bioimageflow_sairpico_tools import MedianDenoising

result = MedianDenoising().process_row(
    Arguments(
        input_image="input.tif",
        denoising_type="2D",
        radius_x=1,
        radius_y=1,
        radius_z=1,
        radius_t=1,
        padding=True,
        output_image="median.tif",
    )
)

Expected Results

The wrapper writes median.tif and returns that path. The example invokes simgmedian2d with -rx 1, -ry 1, and -padding true.

Failure Modes

• The selected simgmedian* command is missing.

• Image dimensionality does not match the selected mode.

• Radius values remove desired structures or create excessive smoothing.

• The subprocess exits non-zero or cannot write the output.