darsia.extract_characteristic_data#

extract_characteristic_data(signal, mask=None, samples=None, filter=<function <lambda>>, num_clusters=5, num_attempts=100, num_iterations=200, eps=0.1, mode='most_common', show_plot=False)[source]#

Extract representative colours from an image for given patches.

Parameters:

signalnumpy.ndarray or darsia.Image: Input signal, assumed to be a 2d colour image.
masknumpy.ndarray or darsia.Image, optional: Boolean array flagging pixels of interest; by default all pixels.
sampleslist of slices, optional: 2d regions of interest; None means the full region.
filtercallable(), optional: Preprocessing applied to the signal before analysis (e.g. a Gaussian filter). Default: identity.
num_clustersint, optional: Number of clusters to extract. Default 5.
num_attemptsint, optional: Number of attempts to find the best clusters. Default 100.
num_iterationsint, optional: Maximum iterations of the iterative procedure. Default 200.
epsfloat, optional: Stopping tolerance for the iterative procedure. Default 0.1.
mode{“most_common”, “least_common”, “all”}, optional: What to return. Default "most_common".
show_plotbool, optional: Display plots. Default False.

Returns:

numpy.ndarray or tuple of numpy.ndarray: Characteristic colours for the chosen samples when mode is "most_common" / "least_common"; a (labels, palettes) tuple when mode is "all".