Classes
class	FactorizedChi2Initialization

class	FactorizedInitialization

class	FactorizedWaveletInitialization

Functions
tuple[np.ndarray, Box]	trim_morphology (np.ndarray morph, float bg_thresh=0, int padding=5)

tuple[Box, np.ndarray\|None]	init_monotonic_morph (np.ndarray detect, tuple[int, int] center, Box full_box, int padding=5, bool normalize=True, Monotonicity\|None monotonicity=None, float thresh=0)

np.ndarray	multifit_spectra (Observation observation, Sequence[Image] morphs, Image\|None model=None)

Variables
	logger = logging.getLogger("scarlet.lite.initialization")

Function Documentation

◆ init_monotonic_morph()

tuple[Box, np.ndarray \| None] lsst.scarlet.lite.initialization.init_monotonic_morph	(	np.ndarray	detect,
		tuple[int, int]	center,
		Box	full_box,
		int	padding = 5,
		bool	normalize = True,
		Monotonicity \| None	monotonicity = None,
		float	thresh = 0 )

Initialize a morphology for a monotonic source

Parameters
----------
detect:
    The 2D detection image contained in `full_box`.
center:
    The center of the monotonic source.
full_box:
    The bounding box of `detect`.
padding:
    The number of pixels to grow the morphology in each direction.
    This can be useful if initializing a source with a kernel that
    is known to be narrower than the expected value of the source.
normalize:
    Whether or not to normalize the morphology.
monotonicity:
    When `monotonicity` is `None`,
    the component is initialized with only the
    monotonic pixels, otherwise the monotonicity operator is used to
    project the morphology to a monotonic solution.
thresh:
    The threshold (fraction above the background) to use for trimming the
    morphology.

Returns
-------
bbox:
    The bounding box of the morphology.
morph:
    The initialized morphology.

Definition at line 70 of file initialization.py.

) -> tuple[Box, np.ndarray | None]:
    """Initialize a morphology for a monotonic source
 
    Parameters
    ----------
    detect:
        The 2D detection image contained in `full_box`.
    center:
        The center of the monotonic source.
    full_box:
        The bounding box of `detect`.
    padding:
        The number of pixels to grow the morphology in each direction.
        This can be useful if initializing a source with a kernel that
        is known to be narrower than the expected value of the source.
    normalize:
        Whether or not to normalize the morphology.
    monotonicity:
        When `monotonicity` is `None`,
        the component is initialized with only the
        monotonic pixels, otherwise the monotonicity operator is used to
        project the morphology to a monotonic solution.
    thresh:
        The threshold (fraction above the background) to use for trimming the
        morphology.
 
    Returns
    -------
    bbox:
        The bounding box of the morphology.
    morph:
        The initialized morphology.
    """
    center: tuple[int, int] = tuple(center[i] - full_box.origin[i] for i in range(2))  # type: ignore
 
    if monotonicity is None:
        _, morph, bounds = prox_monotonic_mask(detect, center, max_iter=0)
        bbox = bounds_to_bbox(bounds)
        if bbox.shape == (1, 1) and morph[bbox.slices][0, 0] == 0:
            return Box((0, 0)), None
 
        if thresh > 0:
            morph, bbox = trim_morphology(morph, bg_thresh=thresh, padding=padding)
 
        # Shift the bounding box to account for the non-zero origin
        bbox += full_box.origin
 
    else:
        morph = monotonicity(detect, center)
 
        # truncate morph at thresh * bg_rms
        morph, bbox = trim_morphology(morph, bg_thresh=thresh, padding=padding)
        # Shift the bounding box to account for the non-zero origin
        bbox += full_box.origin
 
    if np.max(morph) == 0:
        return Box((0, 0), origin=full_box.origin), None
 
    if normalize:
        morph /= np.max(morph)
 
    if padding is not None and padding > 0:
        # Pad the morphology to allow it to grow
        bbox = bbox.grow(padding)
 
    # Ensure that the bounding box is inside the full box,
    # even after padding.
    bbox = bbox & full_box
    return bbox, morph
 
 

◆ multifit_spectra()

np.ndarray lsst.scarlet.lite.initialization.multifit_spectra	(	Observation	observation,
		Sequence[Image]	morphs,
		Image \| None	model = None )

Fit the spectra of multiple components simultaneously

Parameters
----------
observation:
    The class containing the observation data.
morphs:
    The morphology of each component.
model:
    An optional model for sources that are not factorized,
    and thus will not have their spectra fit.
    This model is subtracted from the data before fitting the other
    spectra.

Returns
-------
spectra:
    The spectrum for each component, in the same order as `morphs`.

Definition at line 149 of file initialization.py.

) -> np.ndarray:
    """Fit the spectra of multiple components simultaneously
 
    Parameters
    ----------
    observation:
        The class containing the observation data.
    morphs:
        The morphology of each component.
    model:
        An optional model for sources that are not factorized,
        and thus will not have their spectra fit.
        This model is subtracted from the data before fitting the other
        spectra.
 
    Returns
    -------
    spectra:
        The spectrum for each component, in the same order as `morphs`.
    """
    _bands = observation.bands
    n_bands = len(_bands)
    dtype = observation.images.dtype
 
    if model is not None:
        image = observation.images - model
    else:
        image = observation.images.copy()
 
    morph_images = np.zeros((n_bands, len(morphs), image.data[0].size), dtype=dtype)
    for idx, morph in enumerate(morphs):
        _image = morph.repeat(observation.bands)
        _image = Image.from_box(image.bbox, bands=image.bands).insert(_image)
        morph_images[:, idx] = observation.convolve(_image).data.reshape(n_bands, -1)
 
    spectra = np.zeros((len(morphs), n_bands), dtype=dtype)
 
    for b in range(n_bands):
        a = np.vstack(morph_images[b]).T
        spectra[:, b] = np.linalg.lstsq(a, image[observation.bands[b]].data.flatten(), rcond=None)[0]
    spectra[spectra < 0] = 0
    return spectra
 
 

◆ trim_morphology()

tuple[np.ndarray, Box] lsst.scarlet.lite.initialization.trim_morphology	(	np.ndarray	morph,
		float	bg_thresh = 0,
		int	padding = 5 )

Trim the morphology up to pixels above a threshold

Parameters
----------
morph:
    The morphology to be trimmed.
bg_thresh:
    The morphology is trimmed to pixels above the threshold.
padding:
    The amount to pad each side to allow the source to grow.

Returns
-------
morph:
    The trimmed morphology
box:
    The box that contains the morphology.

Definition at line 40 of file initialization.py.

) -> tuple[np.ndarray, Box]:
    """Trim the morphology up to pixels above a threshold
 
    Parameters
    ----------
    morph:
        The morphology to be trimmed.
    bg_thresh:
        The morphology is trimmed to pixels above the threshold.
    padding:
        The amount to pad each side to allow the source to grow.
 
    Returns
    -------
    morph:
        The trimmed morphology
    box:
        The box that contains the morphology.
    """
    # trim morph to pixels above threshold
    mask = morph > bg_thresh
    morph[~mask] = 0
    bbox = Box.from_data(morph, threshold=0).grow(padding)
    return morph, bbox
 
 

Variable Documentation

◆ logger

lsst.scarlet.lite.initialization.logger = logging.getLogger("scarlet.lite.initialization")

Definition at line 37 of file initialization.py.

Classes