Find Ridges

In this tutorial, learn how to detect density ridges in .fits images and regrid teh results back into image space.

Setup & Detect Ridges

To detect ridges in an image:

from crispy import image_ridge_find as irf

# Set up the input image and parameters
image_file = "input_image.fits"  # Path to the input .fits image
thres = 0.5                      # Minimum density threshold
h = 2                            # Smoothing bandwidth

# Detect ridges
ridge_data = irf.run(image_file, h=h, thres=thres)

Parameter Tips

• h (smoothing bandwidth): controls ridge detection resolution. Use h=2 for Nyquist-sampled images. Higher values reduce noise but lower detail.

• thres (density threshold): minimum intensity for valid pixels. Increase this value for noisier data and faster computing time.

• eps (convergence precision): lower values improve convergence accuracy but may slow down processing.

• Walkers: reduce the number of walkers for faster processing.

Important

Computing time scales nonlinear with the number of walker and image points. Use the thres parameter strategically to minimize the number of image points (i.e., voxels) needed for ridge detection.

Save Results

To save the ridge coordinates to a file:

# Save results to a text file
ridge_file = "ridge_coords.txt"
irf.write_output(coords=ridge_data, fname=ridge_file)

Grid Results

To map the ridge coordinates back onto the native image grid:

from crispy import grid_ridge as gr

# Define file paths
ridge_file = "ridge_coords.txt"     # Ridge coordinates
image_file = "input_image.fits"     # Input reference image
skel_file = "gridded_skel.fits"     # Output skeleton file

# Grid ridge results
gr.grid_skel(readFile=ridge_file, imgFile=image_file, writeFile=skel_file)

The output, gridded_skel.fits, contains the skeletonized ridge map aligned with the input image.