run#

crispy.image_ridge_find.run(image, h=1, eps=0.01, maxT=1000, thres=0.135, ordXYZ=True, crdScaling=None, converge_frac=99, ncpu=None, walkerThres=None, overmask=None, walkers=None, min_size=9, return_unconverged=True, f_h=5)[source]#

Identify density ridges in a gridded image using the SCMS algorithm.

This function serves as a wrapper for the SCMS algorithm to identify density ridges in FITS images or NumPy arrays. It supports configurable smoothing, convergence criteria, and walker initialization for ridge identification.

Parameters:

  • image (str or ndarray) – Input image as a FITS file path or a NumPy array.

  • h (float, optional, default=1) – Smoothing bandwidth of the Gaussian kernel.

  • eps (float, optional, default=1e-2) – Convergence criteria for individual walkers. A smaller value increases precision.

  • maxT (int, optional, default=1000) – Maximum number of iterations allowed for the walkers to converge.

  • thres (float, optional, default=0.135) – Lower intensity threshold. Pixels below this threshold are ignored.

  • ordXYZ (bool, optional, default=True) – If True, reorders coordinates to XYZ format (native Python uses ZYX).

  • crdScaling (list of float, optional) – Scaling factors for each axis. Coordinates are scaled during processing and rescaled in the output.

  • converge_frac (float, optional, default=99) – Fraction of walkers that need to converge for the process to terminate, expressed as a percentage.

  • ncpu (int, optional) – Number of CPUs to use. If None, defaults to one less than the total number of CPUs.

  • walkerThres (float, optional) – Lower intensity threshold for placing walkers. Overrides the automated walker placement if provided.

  • overmask (ndarray of bool, optional) – Boolean mask specifying which pixels to include in ridge finding, in addition to the thres criteria.

  • walkers (ndarray, optional) – Custom initial positions for walkers. Overrides automated walker placement if provided.

  • min_size (int, optional, default=9) – Minimum size (in pixels) for structures to be considered. Smaller structures are removed.

  • return_unconverged (bool, optional, default=True) – If True, includes both converged and unconverged walkers in the output. Otherwise, only converged walkers are returned.

  • f_h (float, optional, default=5) – Filtering factor used to exclude data points based on their distance to all other points.

Returns:

Coordinates of the density ridges. If return_unconverged is True, returns a tuple: (converged walkers, unconverged walkers). Otherwise, only the converged walkers are returned.

Return type:

ndarray or tuple of ndarray

Notes

  • The image can be a 2D or 3D array. For 3D images, smoothing and walker placement apply to all axes equally unless crdScaling is provided.

  • If crdScaling is used, the coordinates are scaled back to their original values in the output.

  • This function is designed to work efficiently with gridded images and leverages multiprocessing.

Examples

Process a FITS image and retrieve density ridges:

>>> from crispy import image_ridge_find as irf
>>> ridges = irf.run("input_image.fits", h=2, thres=0.2, maxT=500)

Use custom walkers for initialization:

>>> walkers = np.array([[10, 20], [15, 25], [20, 30]])
>>> ridges = irf.run("input_image.fits", walkers=walkers, maxT=200)
On this page

This Page