Axis finder

This is a reimplementation of the axis finder provided in the Matlab reference implementation for the seminar on tomography at the University of Vienna (winter semester 2025/26). It is built using the dashboarding library panel.

Usage

To use the axis finder, you can first install the required packages (ideally in a virtual environment) by running:

pip install -e .

Then the axis finder can be started with:

axis-finder path/to/image/directory

This will open the GUI in a browser window. To auto-reload the applet if you make changes to the code, you can run axis-finder with the --dev flag:

axis-finder path/to/image/directory --dev

In a Jupyter Notebook

The axis finder can also be used in a Jupyter notebook. To that end, first load the required modules and make sure to enable the panel extension:

import panel as pn
from axis_finder.app import App as AxisFinder

pn.extension()

The the axis finder object can be created and displayed as follows:

app = AxisFinder("path/to/image/directory")
app.build()

API Reference

In the notebook, one can then also access the center position and crop dimensions as well as the image data directly via the app object.

Properties

These properties map directly to the panel slider values.

Property	Return Type	Description
cx	float	Current x-coordinate center for the crop area.
cy	float	Current y-coordinate center for the crop area.
crop_width	float/int	The width of the bounding box to be cropped.
crop_height	float/int	The height of the bounding box to be cropped.
first_image_index	int	The 0-based start index calculated from the image range slider.
image_count	int	The total number of images included in the selected range.

---

Methods

iter_image_paths()

A generator that yields the absolute paths of the images in the currently selected range.

• Yields: str (Path to image file)

get_grayscale_cropped_image(index)

Fetches a single image, applies the crop defined by the sliders, and returns it in grayscale.

• Arguments: index (int) - The relative index within the current selection.
• Returns: np.ndarray - The cropped grayscale image data.

iter_cropped_grayscale_images()

A lazy generator that iterates through the entire selected range and yields processed images.

• Yields: np.ndarray - Processed image data for batch analysis.

build()

Constructs the Panel UI layout.

• Returns: The layout containing the sliders and controls.

---

Assuming the app object has been created as shown above, the properties and methods can be used as follows:

# Access properties
center_x = app.cx
center_y = app.cy

print(f"Center X: {center_x}, Center Y: {center_y}")

# Iterate over image paths
for path in app.iter_image_paths():
    print(path)

# Get a single cropped grayscale image
image = app.get_grayscale_cropped_image(0)
print(image.shape)

An example notebook is also provided in examples/example_notebook.ipynb.

Docker/Podman

The app can also be bundled in a Docker or Podman container. To build the container image, run:

docker build -t axis-finder .

or with Podman:

podman build -t axis-finder .

To run the container, use:

docker run --rm -p 5006:5006 -v path/to/image/directory:/data axis-finder /data

or with Podman:

podman run --rm -p 5006:5006 -v path/to/image/directory:/data axis-finder /data