lsst.pipe.tasks.prettyPictureMaker._functors._gamut_fixer Namespace Reference

Classes
class	GamutFixer

Functions
RGBImage	heal_gamut (LABImage lab_image, np.ndarray[tuple[int, int], np.bool] mask, tuple[float, float] xyz_whitepoint, int max_size=500, int dilation_iterations=3)

Function Documentation

heal_gamut()

RGBImage heal_gamut	(	LABImage	lab_image,
		np.ndarray[tuple[int, int], np.bool]	mask,
		tuple[float, float]	xyz_whitepoint,
		int	max_size = 500,
		int	dilation_iterations = 3 )

Heal out-of-gamut regions in a Lab image using diffusion-based inpainting.

Parameters
----------
lab_image : `LABImage`
    A NxMx3 array in the Lab colorspace. Modified in-place by healing out-of-gamut regions.
    For each region, a copy is made before modification, then written back.
mask : `numpy.ndarray` of `bool`
    A boolean mask indicating which pixels are out of gamut.
xyz_whitepoint : `tuple` of `float`, `float`
    Sets the white point of the xyz colorspace in xy coordinates.
max_size : `int`, optional
    Maximum size of regions to heal. Larger regions are skipped. Default is 500.
dilation_iterations : `int`, optional
    Number of iterations for mask dilation. Default is 3.

Returns
-------
result : `RGBImage`
    The healed image converted to RGB colorspace.

Raises
------
ValueError
    Raised if the shapes of lab_image and mask are incompatible.

Notes
-----
The healing algorithm works by:

1. Labeling connected regions in the mask
2. For each region smaller than max_size:

   - Dilating the mask to create an annulus around the region
   - Computing average a,b color values from the annulus
   - Using rgb.inpaint_mask to interpolate the L, a, and b channels
   - Filling the masked region with the interpolated values

3. Regions larger than max_size are skipped.

Definition at line 39 of file _gamut_fixer.py.