scikit-image

0.26.0 · active · verified Sun Mar 29

scikit-image is an open-source Python library for image processing and computer vision, built on NumPy, SciPy, and other core scientific Python projects. It provides a comprehensive collection of algorithms for tasks such as segmentation, filtering, transformation, feature detection, and analysis. The project is actively maintained, with frequent releases, and version 0.26.0 was released on 2025-12-20.

Warnings

deprecated The binary versions of morphological operations (`binary_erosion`, `binary_dilation`, `binary_opening`, `binary_closing`) in `skimage.morphology` are deprecated in favor of their non-binary counterparts (`erosion`, `dilation`, `opening`, `closing`). The binary versions were not significantly faster and sometimes slower.
Fix: Replace calls to `skimage.morphology.binary_erosion` with `skimage.morphology.erosion`, `binary_dilation` with `dilation`, etc.
deprecated Parameters `max_cost` and `max_cumulative_cost` in `skimage.graph.MCP.find_costs` are deprecated as they previously did nothing. A new parameter `max_step_cost` has been introduced.
Fix: Remove `max_cost` and `max_cumulative_cost` arguments. If limiting step cost is desired, use the new `max_step_cost` parameter.
deprecated The `area_threshold` parameter in `skimage.morphology.remove_small_holes` is deprecated in favor of `max_size` for clearer API and behavior. `max_size` removes holes smaller than or equal to its value, while the previous parameter only removed strictly smaller ones.
Fix: Update calls to `skimage.morphology.remove_small_holes` by replacing `area_threshold` with `max_size`, adjusting the value if necessary to match the desired behavior.
gotcha scikit-image 0.26.0 requires Python 3.11 or newer. Attempting to install with an older Python version will automatically resolve to an older, compatible `scikit-image` version, which may lack recent features and fixes.
Fix: Ensure your Python environment is at least 3.11 (preferably 3.12+). Use a virtual environment to manage Python versions and dependencies.

Install

pip install scikit-image Install core library
pip install scikit-image[data] Install with example datasets (requires pooch)
pip install scikit-image[optional] Install with optional scientific packages (e.g., for parallel processing)

Imports

skimage
```
import skimage as ski
```
The standard and recommended import alias for scikit-image.
io
```
from skimage import io
```
Commonly imported submodule for image input/output operations like imread/imwrite.
filters
```
from skimage import filters
```
Commonly imported submodule for image filtering operations.

Quickstart

This quickstart demonstrates loading an example grayscale image using `skimage.data.coins()` and applying a common image filtering technique, Otsu thresholding, from `skimage.filters` to convert it into a binary image. The shapes of the original and processed images are printed to confirm the operations.

import skimage as ski

# Load an example image (e.g., coins dataset)
image = ski.data.coins()

# Apply a basic image processing operation, e.g., Otsu thresholding
threshold_value = ski.filters.threshold_otsu(image)
binary_image = image > threshold_value

print(f"Original image shape: {image.shape}")
print(f"Calculated threshold value: {threshold_value}")
print(f"Processed (binary) image shape: {binary_image.shape}")

view raw JSON →