lsst.ip.diffim.dcrModel.DcrModel Class Reference

Public Member Functions
def	__init__ (self, modelImages, effectiveWavelength, bandwidth, filterLabel=None, psf=None, mask=None, variance=None, photoCalib=None)
def	fromImage (cls, maskedImage, dcrNumSubfilters, effectiveWavelength, bandwidth, filterLabel=None, psf=None, photoCalib=None)
def	fromDataRef (cls, dataRef, effectiveWavelength, bandwidth, datasetType="dcrCoadd", numSubfilters=None, **kwargs)
def	fromQuantum (cls, availableCoaddRefs, effectiveWavelength, bandwidth)
def	__len__ (self)
def	__getitem__ (self, subfilter)
def	__setitem__ (self, subfilter, maskedImage)
def	effectiveWavelength (self)
def	filter (self)
def	bandwidth (self)
def	psf (self)
def	bbox (self)
def	mask (self)
def	variance (self)
def	getReferenceImage (self, bbox=None)
def	assign (self, dcrSubModel, bbox=None)
def	buildMatchedTemplate (self, exposure=None, order=3, visitInfo=None, bbox=None, wcs=None, mask=None, splitSubfilters=True, splitThreshold=0., amplifyModel=1.)
def	buildMatchedExposure (self, exposure=None, visitInfo=None, bbox=None, wcs=None, mask=None)
def	conditionDcrModel (self, modelImages, bbox, gain=1.)
def	regularizeModelIter (self, subfilter, newModel, bbox, regularizationFactor, regularizationWidth=2)
def	regularizeModelFreq (self, modelImages, bbox, statsCtrl, regularizationFactor, regularizationWidth=2, mask=None, convergenceMaskPlanes="DETECTED")
def	calculateNoiseCutoff (self, image, statsCtrl, bufferSize, convergenceMaskPlanes="DETECTED", mask=None, bbox=None)
def	applyImageThresholds (self, image, highThreshold=None, lowThreshold=None, regularizationWidth=2)

Public Attributes
	dcrNumSubfilters
	modelImages
	photoCalib

Detailed Description

A model of the true sky after correcting chromatic effects.

Attributes
----------
dcrNumSubfilters : `int`
    Number of sub-filters used to model chromatic effects within a band.
modelImages : `list` of `lsst.afw.image.Image`
    A list of masked images, each containing the model for one subfilter

Notes
-----
The ``DcrModel`` contains an estimate of the true sky, at a higher
wavelength resolution than the input observations. It can be forward-
modeled to produce Differential Chromatic Refraction (DCR) matched
templates for a given ``Exposure``, and provides utilities for conditioning
the model in ``dcrAssembleCoadd`` to avoid oscillating solutions between
iterations of forward modeling or between the subfilters of the model.

Definition at line 32 of file dcrModel.py.

Constructor & Destructor Documentation

__init__()

def lsst.ip.diffim.dcrModel.DcrModel.__init__	(		self,
			modelImages,
			effectiveWavelength,
			bandwidth,
			filterLabel = None,
			psf = None,
			mask = None,
			variance = None,
			photoCalib = None
	)

Definition at line 52 of file dcrModel.py.

                 mask=None, variance=None, photoCalib=None):
        self.dcrNumSubfilters = len(modelImages)
        self.modelImages = modelImages
        self._filterLabel = filterLabel
        self._effectiveWavelength = effectiveWavelength
        self._bandwidth = bandwidth
        self._psf = psf
        self._mask = mask
        self._variance = variance
        self.photoCalib = photoCalib
 

Member Function Documentation

__getitem__()

def lsst.ip.diffim.dcrModel.DcrModel.__getitem__	(		self,
			subfilter
	)

Iterate over the subfilters of the DCR model.

Parameters
----------
subfilter : `int`
    Index of the current ``subfilter`` within the full band.
    Negative indices are allowed, and count in reverse order
    from the highest ``subfilter``.

Returns
-------
modelImage : `lsst.afw.image.Image`
    The DCR model for the given ``subfilter``.

Raises
------
IndexError
    If the requested ``subfilter`` is greater or equal to the number
    of subfilters in the model.

Definition at line 220 of file dcrModel.py.

    def __getitem__(self, subfilter):
        """Iterate over the subfilters of the DCR model.
 
        Parameters
        ----------
        subfilter : `int`
            Index of the current ``subfilter`` within the full band.
            Negative indices are allowed, and count in reverse order
            from the highest ``subfilter``.
 
        Returns
        -------
        modelImage : `lsst.afw.image.Image`
            The DCR model for the given ``subfilter``.
 
        Raises
        ------
        IndexError
            If the requested ``subfilter`` is greater or equal to the number
            of subfilters in the model.
        """
        if np.abs(subfilter) >= len(self):
            raise IndexError("subfilter out of bounds.")
        return self.modelImages[subfilter]
 

__len__()

def lsst.ip.diffim.dcrModel.DcrModel.__len__

(

self

)

Return the number of subfilters.

Returns
-------
dcrNumSubfilters : `int`
    The number of DCR subfilters in the model.

Definition at line 210 of file dcrModel.py.

    def __len__(self):
        """Return the number of subfilters.
 
        Returns
        -------
        dcrNumSubfilters : `int`
            The number of DCR subfilters in the model.
        """
        return self.dcrNumSubfilters
 

__setitem__()

def lsst.ip.diffim.dcrModel.DcrModel.__setitem__	(		self,
			subfilter,
			maskedImage
	)

Update the model image for one subfilter.

Parameters
----------
subfilter : `int`
    Index of the current subfilter within the full band.
maskedImage : `lsst.afw.image.Image`
    The DCR model to set for the given ``subfilter``.

Raises
------
IndexError
    If the requested ``subfilter`` is greater or equal to the number
    of subfilters in the model.
ValueError
    If the bounding box of the new image does not match.

Definition at line 245 of file dcrModel.py.

    def __setitem__(self, subfilter, maskedImage):
        """Update the model image for one subfilter.
 
        Parameters
        ----------
        subfilter : `int`
            Index of the current subfilter within the full band.
        maskedImage : `lsst.afw.image.Image`
            The DCR model to set for the given ``subfilter``.
 
        Raises
        ------
        IndexError
            If the requested ``subfilter`` is greater or equal to the number
            of subfilters in the model.
        ValueError
            If the bounding box of the new image does not match.
        """
        if np.abs(subfilter) >= len(self):
            raise IndexError("subfilter out of bounds.")
        if maskedImage.getBBox() != self.bbox:
            raise ValueError("The bounding box of a subfilter must not change.")
        self.modelImages[subfilter] = maskedImage
 

applyImageThresholds()

def lsst.ip.diffim.dcrModel.DcrModel.applyImageThresholds	(		self,
			image,
			highThreshold = None,
			lowThreshold = None,
			regularizationWidth = 2
	)

Restrict image values to be between upper and lower limits.

This method flags all pixels in an image that are outside of the given
threshold values. The threshold values are taken from a reference
image, so noisy pixels are likely to get flagged. In order to exclude
those noisy pixels, the array of flags is eroded and dilated, which
removes isolated pixels outside of the thresholds from the list of
pixels to be modified. Pixels that remain flagged after this operation
have their values set to the appropriate upper or lower threshold
value.

Parameters
----------
image : `numpy.ndarray`
    The image to apply the thresholds to.
    The values will be modified in place.
highThreshold : `numpy.ndarray`, optional
    Array of upper limit values for each pixel of ``image``.
lowThreshold : `numpy.ndarray`, optional
    Array of lower limit values for each pixel of ``image``.
regularizationWidth : `int`, optional
    Minimum radius of a region to include in regularization, in pixels.

Definition at line 650 of file dcrModel.py.

    def applyImageThresholds(self, image, highThreshold=None, lowThreshold=None, regularizationWidth=2):
        """Restrict image values to be between upper and lower limits.
 
        This method flags all pixels in an image that are outside of the given
        threshold values. The threshold values are taken from a reference
        image, so noisy pixels are likely to get flagged. In order to exclude
        those noisy pixels, the array of flags is eroded and dilated, which
        removes isolated pixels outside of the thresholds from the list of
        pixels to be modified. Pixels that remain flagged after this operation
        have their values set to the appropriate upper or lower threshold
        value.
 
        Parameters
        ----------
        image : `numpy.ndarray`
            The image to apply the thresholds to.
            The values will be modified in place.
        highThreshold : `numpy.ndarray`, optional
            Array of upper limit values for each pixel of ``image``.
        lowThreshold : `numpy.ndarray`, optional
            Array of lower limit values for each pixel of ``image``.
        regularizationWidth : `int`, optional
            Minimum radius of a region to include in regularization, in pixels.
        """
        # Generate the structure for binary erosion and dilation, which is used
        # to remove noise-like pixels. Groups of pixels with a radius smaller
        # than ``regularizationWidth`` will be excluded from regularization.
        filterStructure = ndimage.iterate_structure(ndimage.generate_binary_structure(2, 1),
                                                    regularizationWidth)
        if highThreshold is not None:
            highPixels = image > highThreshold
            if regularizationWidth > 0:
                # Erode and dilate ``highPixels`` to exclude noisy pixels.
                highPixels = ndimage.morphology.binary_opening(highPixels, structure=filterStructure)
            image[highPixels] = highThreshold[highPixels]
        if lowThreshold is not None:
            lowPixels = image < lowThreshold
            if regularizationWidth > 0:
                # Erode and dilate ``lowPixels`` to exclude noisy pixels.
                lowPixels = ndimage.morphology.binary_opening(lowPixels, structure=filterStructure)
            image[lowPixels] = lowThreshold[lowPixels]
 
 

assign()

def lsst.ip.diffim.dcrModel.DcrModel.assign	(		self,
			dcrSubModel,
			bbox = None
	)

Update a sub-region of the ``DcrModel`` with new values.

Parameters
----------
dcrSubModel : `lsst.pipe.tasks.DcrModel`
    New model of the true scene after correcting chromatic effects.
bbox : `lsst.afw.geom.Box2I`, optional
    Sub-region of the coadd.
    Defaults to the bounding box of ``dcrSubModel``.

Raises
------
ValueError
    If the new model has a different number of subfilters.

Definition at line 363 of file dcrModel.py.

    def assign(self, dcrSubModel, bbox=None):
        """Update a sub-region of the ``DcrModel`` with new values.
 
        Parameters
        ----------
        dcrSubModel : `lsst.pipe.tasks.DcrModel`
            New model of the true scene after correcting chromatic effects.
        bbox : `lsst.afw.geom.Box2I`, optional
            Sub-region of the coadd.
            Defaults to the bounding box of ``dcrSubModel``.
 
        Raises
        ------
        ValueError
            If the new model has a different number of subfilters.
        """
        if len(dcrSubModel) != len(self):
            raise ValueError("The number of DCR subfilters must be the same "
                             "between the old and new models.")
        bbox = bbox or self.bbox
        for model, subModel in zip(self, dcrSubModel):
            model.assign(subModel[bbox], bbox)
 

bandwidth()

def lsst.ip.diffim.dcrModel.DcrModel.bandwidth

(

self

)

Return the bandwidth of the model.

Returns
-------
bandwidth : `float`
    The bandwidth of the current filter, in nanometers.

Definition at line 292 of file dcrModel.py.

    def bandwidth(self):
        """Return the bandwidth of the model.
 
        Returns
        -------
        bandwidth : `float`
            The bandwidth of the current filter, in nanometers.
        """
        return self._bandwidth
 

bbox()

def lsst.ip.diffim.dcrModel.DcrModel.bbox

(

self

)

Return the common bounding box of each subfilter image.

Returns
-------
bbox : `lsst.afw.geom.Box2I`
    Bounding box of the DCR model.

Definition at line 314 of file dcrModel.py.

    def bbox(self):
        """Return the common bounding box of each subfilter image.
 
        Returns
        -------
        bbox : `lsst.afw.geom.Box2I`
            Bounding box of the DCR model.
        """
        return self[0].getBBox()
 

buildMatchedExposure()

def lsst.ip.diffim.dcrModel.DcrModel.buildMatchedExposure	(		self,
			exposure = None,
			visitInfo = None,
			bbox = None,
			wcs = None,
			mask = None
	)

Wrapper to create an exposure from a template image.

Parameters
----------
exposure : `lsst.afw.image.Exposure`, optional
    The input exposure to build a matched template for.
    May be omitted if all of the metadata is supplied separately
visitInfo : `lsst.afw.image.VisitInfo`, optional
    Metadata for the exposure. Ignored if ``exposure`` is set.
bbox : `lsst.afw.geom.Box2I`, optional
    Sub-region of the coadd. Ignored if ``exposure`` is set.
wcs : `lsst.afw.geom.SkyWcs`, optional
    Coordinate system definition (wcs) for the exposure.
    Ignored if ``exposure`` is set.
mask : `lsst.afw.image.Mask`, optional
    reference mask to use for the template image.

Returns
-------
templateExposure : `lsst.afw.image.exposureF`
    The DCR-matched template

Raises
------
RuntimeError
    If no `photcCalib` is set.

Definition at line 450 of file dcrModel.py.

                             visitInfo=None, bbox=None, wcs=None, mask=None):
        """Wrapper to create an exposure from a template image.
 
        Parameters
        ----------
        exposure : `lsst.afw.image.Exposure`, optional
            The input exposure to build a matched template for.
            May be omitted if all of the metadata is supplied separately
        visitInfo : `lsst.afw.image.VisitInfo`, optional
            Metadata for the exposure. Ignored if ``exposure`` is set.
        bbox : `lsst.afw.geom.Box2I`, optional
            Sub-region of the coadd. Ignored if ``exposure`` is set.
        wcs : `lsst.afw.geom.SkyWcs`, optional
            Coordinate system definition (wcs) for the exposure.
            Ignored if ``exposure`` is set.
        mask : `lsst.afw.image.Mask`, optional
            reference mask to use for the template image.
 
        Returns
        -------
        templateExposure : `lsst.afw.image.exposureF`
            The DCR-matched template
 
        Raises
        ------
        RuntimeError
            If no `photcCalib` is set.
        """
        if bbox is None:
            bbox = exposure.getBBox()
        templateImage = self.buildMatchedTemplate(exposure=exposure, visitInfo=visitInfo,
                                                  bbox=bbox, wcs=wcs, mask=mask)
        maskedImage = afwImage.MaskedImageF(bbox)
        maskedImage.image = templateImage[bbox]
        maskedImage.mask = self.mask[bbox]
        maskedImage.variance = self.variance[bbox]
        # The variance of the stacked image will be `dcrNumSubfilters`
        # times the variance of the individual subfilters.
        maskedImage.variance *= self.dcrNumSubfilters
        templateExposure = afwImage.ExposureF(bbox, wcs)
        templateExposure.setMaskedImage(maskedImage[bbox])
        templateExposure.setPsf(self.psf)
        templateExposure.setFilterLabel(self.filterLabel)
        if self.photoCalib is None:
            raise RuntimeError("No PhotoCalib set for the DcrModel. "
                               "If the DcrModel was created from a masked image"
                               " you must also specify the photoCalib.")
        templateExposure.setPhotoCalib(self.photoCalib)
        return templateExposure
 

buildMatchedTemplate()

def lsst.ip.diffim.dcrModel.DcrModel.buildMatchedTemplate	(		self,
			exposure = None,
			order = 3,
			visitInfo = None,
			bbox = None,
			wcs = None,
			mask = None,
			splitSubfilters = True,
			splitThreshold = 0.,
			amplifyModel = 1.
	)

Create a DCR-matched template image for an exposure.

Parameters
----------
exposure : `lsst.afw.image.Exposure`, optional
    The input exposure to build a matched template for.
    May be omitted if all of the metadata is supplied separately
order : `int`, optional
    Interpolation order of the DCR shift.
visitInfo : `lsst.afw.image.VisitInfo`, optional
    Metadata for the exposure. Ignored if ``exposure`` is set.
bbox : `lsst.afw.geom.Box2I`, optional
    Sub-region of the coadd. Ignored if ``exposure`` is set.
wcs : `lsst.afw.geom.SkyWcs`, optional
    Coordinate system definition (wcs) for the exposure.
    Ignored if ``exposure`` is set.
mask : `lsst.afw.image.Mask`, optional
    reference mask to use for the template image.
splitSubfilters : `bool`, optional
    Calculate DCR for two evenly-spaced wavelengths in each subfilter,
    instead of at the midpoint. Default: True
splitThreshold : `float`, optional
    Minimum DCR difference within a subfilter required to use
    ``splitSubfilters``
amplifyModel : `float`, optional
    Multiplication factor to amplify differences between model planes.
    Used to speed convergence of iterative forward modeling.

Returns
-------
templateImage : `lsst.afw.image.ImageF`
    The DCR-matched template

Raises
------
ValueError
    If neither ``exposure`` or all of ``visitInfo``, ``bbox``, and
    ``wcs`` are set.

Definition at line 386 of file dcrModel.py.

                             splitSubfilters=True, splitThreshold=0., amplifyModel=1.):
        """Create a DCR-matched template image for an exposure.
 
        Parameters
        ----------
        exposure : `lsst.afw.image.Exposure`, optional
            The input exposure to build a matched template for.
            May be omitted if all of the metadata is supplied separately
        order : `int`, optional
            Interpolation order of the DCR shift.
        visitInfo : `lsst.afw.image.VisitInfo`, optional
            Metadata for the exposure. Ignored if ``exposure`` is set.
        bbox : `lsst.afw.geom.Box2I`, optional
            Sub-region of the coadd. Ignored if ``exposure`` is set.
        wcs : `lsst.afw.geom.SkyWcs`, optional
            Coordinate system definition (wcs) for the exposure.
            Ignored if ``exposure`` is set.
        mask : `lsst.afw.image.Mask`, optional
            reference mask to use for the template image.
        splitSubfilters : `bool`, optional
            Calculate DCR for two evenly-spaced wavelengths in each subfilter,
            instead of at the midpoint. Default: True
        splitThreshold : `float`, optional
            Minimum DCR difference within a subfilter required to use
            ``splitSubfilters``
        amplifyModel : `float`, optional
            Multiplication factor to amplify differences between model planes.
            Used to speed convergence of iterative forward modeling.
 
        Returns
        -------
        templateImage : `lsst.afw.image.ImageF`
            The DCR-matched template
 
        Raises
        ------
        ValueError
            If neither ``exposure`` or all of ``visitInfo``, ``bbox``, and
            ``wcs`` are set.
        """
        if self.effectiveWavelength is None or self.bandwidth is None:
            raise ValueError("'effectiveWavelength' and 'bandwidth' must be set for the DcrModel in order "
                             "to calculate DCR.")
        if exposure is not None:
            visitInfo = exposure.getInfo().getVisitInfo()
            bbox = exposure.getBBox()
            wcs = exposure.getInfo().getWcs()
        elif visitInfo is None or bbox is None or wcs is None:
            raise ValueError("Either exposure or visitInfo, bbox, and wcs must be set.")
        dcrShift = calculateDcr(visitInfo, wcs, self.effectiveWavelength, self.bandwidth, len(self),
                                splitSubfilters=splitSubfilters)
        templateImage = afwImage.ImageF(bbox)
        refModel = self.getReferenceImage(bbox)
        for subfilter, dcr in enumerate(dcrShift):
            if amplifyModel > 1:
                model = (self[subfilter][bbox].array - refModel)*amplifyModel + refModel
            else:
                model = self[subfilter][bbox].array
            templateImage.array += applyDcr(model, dcr, splitSubfilters=splitSubfilters,
                                            splitThreshold=splitThreshold, order=order)
        return templateImage
 

calculateNoiseCutoff()

def lsst.ip.diffim.dcrModel.DcrModel.calculateNoiseCutoff	(		self,
			image,
			statsCtrl,
			bufferSize,
			convergenceMaskPlanes = "DETECTED",
			mask = None,
			bbox = None
	)

Helper function to calculate the background noise level of an image.

Parameters
----------
image : `lsst.afw.image.Image`
    The input image to evaluate the background noise properties.
statsCtrl : `lsst.afw.math.StatisticsControl`
    Statistics control object for coaddition.
bufferSize : `int`
    Number of additional pixels to exclude
    from the edges of the bounding box.
convergenceMaskPlanes : `list` of `str`, or `str`
    Mask planes to use to calculate convergence.
mask : `lsst.afw.image.Mask`, Optional
    Optional alternate mask
bbox : `lsst.afw.geom.Box2I`, optional
    Sub-region of the masked image to calculate the noise level over.

Returns
-------
noiseCutoff : `float`
    The threshold value to treat pixels as noise in an image..

Definition at line 613 of file dcrModel.py.

                             convergenceMaskPlanes="DETECTED", mask=None, bbox=None):
        """Helper function to calculate the background noise level of an image.
 
        Parameters
        ----------
        image : `lsst.afw.image.Image`
            The input image to evaluate the background noise properties.
        statsCtrl : `lsst.afw.math.StatisticsControl`
            Statistics control object for coaddition.
        bufferSize : `int`
            Number of additional pixels to exclude
            from the edges of the bounding box.
        convergenceMaskPlanes : `list` of `str`, or `str`
            Mask planes to use to calculate convergence.
        mask : `lsst.afw.image.Mask`, Optional
            Optional alternate mask
        bbox : `lsst.afw.geom.Box2I`, optional
            Sub-region of the masked image to calculate the noise level over.
 
        Returns
        -------
        noiseCutoff : `float`
            The threshold value to treat pixels as noise in an image..
        """
        if bbox is None:
            bbox = self.bbox
        if mask is None:
            mask = self.mask[bbox]
        bboxShrink = geom.Box2I(bbox)
        bboxShrink.grow(-bufferSize)
        convergeMask = mask.getPlaneBitMask(convergenceMaskPlanes)
 
        backgroundPixels = mask[bboxShrink].array & (statsCtrl.getAndMask() | convergeMask) == 0
        noiseCutoff = np.std(image[bboxShrink].array[backgroundPixels])
        return noiseCutoff
 

conditionDcrModel()

def lsst.ip.diffim.dcrModel.DcrModel.conditionDcrModel	(		self,
			modelImages,
			bbox,
			gain = 1.
	)

Average two iterations' solutions to reduce oscillations.

Parameters
----------
modelImages : `list` of `lsst.afw.image.Image`
    The new DCR model images from the current iteration.
    The values will be modified in place.
bbox : `lsst.afw.geom.Box2I`
    Sub-region of the coadd
gain : `float`, optional
    Relative weight to give the new solution when updating the model.
    Defaults to 1.0, which gives equal weight to both solutions.

Definition at line 501 of file dcrModel.py.

    def conditionDcrModel(self, modelImages, bbox, gain=1.):
        """Average two iterations' solutions to reduce oscillations.
 
        Parameters
        ----------
        modelImages : `list` of `lsst.afw.image.Image`
            The new DCR model images from the current iteration.
            The values will be modified in place.
        bbox : `lsst.afw.geom.Box2I`
            Sub-region of the coadd
        gain : `float`, optional
            Relative weight to give the new solution when updating the model.
            Defaults to 1.0, which gives equal weight to both solutions.
        """
        # Calculate weighted averages of the images.
        for model, newModel in zip(self, modelImages):
            newModel *= gain
            newModel += model[bbox]
            newModel /= 1. + gain
 

effectiveWavelength()

def lsst.ip.diffim.dcrModel.DcrModel.effectiveWavelength

(

self

)

Return the effective wavelength of the model.

Returns
-------
effectiveWavelength : `float`
    The effective wavelength of the current filter, in nanometers.

Definition at line 270 of file dcrModel.py.

    def effectiveWavelength(self):
        """Return the effective wavelength of the model.
 
        Returns
        -------
        effectiveWavelength : `float`
            The effective wavelength of the current filter, in nanometers.
        """
        return self._effectiveWavelength
 

filter()

def lsst.ip.diffim.dcrModel.DcrModel.filter

(

self

)

Return the filter label for the model.

Returns
-------
filterLabel : `lsst.afw.image.FilterLabel`
    The filter used for the input observations.

Definition at line 281 of file dcrModel.py.

    def filter(self):
        """Return the filter label for the model.
 
        Returns
        -------
        filterLabel : `lsst.afw.image.FilterLabel`
            The filter used for the input observations.
        """
        return self._filterLabel
 

fromDataRef()

def lsst.ip.diffim.dcrModel.DcrModel.fromDataRef	(		cls,
			dataRef,
			effectiveWavelength,
			bandwidth,
			datasetType = "dcrCoadd",
			numSubfilters = None,
		**	kwargs
	)

Load an existing DcrModel from a Gen 2 repository.

Parameters
----------
dataRef : `lsst.daf.persistence.ButlerDataRef`
    Data reference defining the patch for coaddition and the
    reference Warp
effectiveWavelength : `float`
    The effective wavelengths of the current filter, in nanometers.
bandwidth : `float`
    The bandwidth of the current filter, in nanometers.
datasetType : `str`, optional
    Name of the DcrModel in the registry {"dcrCoadd", "dcrCoadd_sub"}
numSubfilters : `int`
    Number of sub-filters used to model chromatic effects within a
    band.
**kwargs
    Additional keyword arguments to pass to look up the model in the
    data registry.
    Common keywords and their types include: ``tract``:`str`,
    ``patch``:`str`, ``bbox``:`lsst.afw.geom.Box2I`

Returns
-------
dcrModel : `lsst.pipe.tasks.DcrModel`
    Best fit model of the true sky after correcting chromatic effects.

Definition at line 115 of file dcrModel.py.

                    **kwargs):
        """Load an existing DcrModel from a Gen 2 repository.
 
        Parameters
        ----------
        dataRef : `lsst.daf.persistence.ButlerDataRef`
            Data reference defining the patch for coaddition and the
            reference Warp
        effectiveWavelength : `float`
            The effective wavelengths of the current filter, in nanometers.
        bandwidth : `float`
            The bandwidth of the current filter, in nanometers.
        datasetType : `str`, optional
            Name of the DcrModel in the registry {"dcrCoadd", "dcrCoadd_sub"}
        numSubfilters : `int`
            Number of sub-filters used to model chromatic effects within a
            band.
        **kwargs
            Additional keyword arguments to pass to look up the model in the
            data registry.
            Common keywords and their types include: ``tract``:`str`,
            ``patch``:`str`, ``bbox``:`lsst.afw.geom.Box2I`
 
        Returns
        -------
        dcrModel : `lsst.pipe.tasks.DcrModel`
            Best fit model of the true sky after correcting chromatic effects.
        """
        modelImages = []
        filterLabel = None
        psf = None
        mask = None
        variance = None
        photoCalib = None
        if "subfilter" in kwargs:
            kwargs.pop("subfilter")
        for subfilter in range(numSubfilters):
            dcrCoadd = dataRef.get(datasetType, subfilter=subfilter,
                                   numSubfilters=numSubfilters, **kwargs)
            if filterLabel is None:
                filterLabel = dcrCoadd.getFilterLabel()
            if psf is None:
                psf = dcrCoadd.getPsf()
            if mask is None:
                mask = dcrCoadd.mask
            if variance is None:
                variance = dcrCoadd.variance
            if photoCalib is None:
                photoCalib = dcrCoadd.getPhotoCalib()
            modelImages.append(dcrCoadd.image)
        return cls(modelImages, effectiveWavelength, bandwidth, filterLabel, psf, mask, variance, photoCalib)
 

fromImage()

def lsst.ip.diffim.dcrModel.DcrModel.fromImage	(		cls,
			maskedImage,
			dcrNumSubfilters,
			effectiveWavelength,
			bandwidth,
			filterLabel = None,
			psf = None,
			photoCalib = None
	)

Initialize a DcrModel by dividing a coadd between the subfilters.

Parameters
----------
maskedImage : `lsst.afw.image.MaskedImage`
    Input coadded image to divide equally between the subfilters.
dcrNumSubfilters : `int`
    Number of sub-filters used to model chromatic effects within a
    band.
effectiveWavelength : `float`
    The effective wavelengths of the current filter, in nanometers.
bandwidth : `float`
    The bandwidth of the current filter, in nanometers.
filterLabel : `lsst.afw.image.FilterLabel`, optional
    The filter label, set in the current instruments' obs package.
    Required for any calculation of DCR, including making matched
    templates.
psf : `lsst.afw.detection.Psf`, optional
    Point spread function (PSF) of the model.
    Required if the ``DcrModel`` will be persisted.
photoCalib : `lsst.afw.image.PhotoCalib`, optional
    Calibration to convert instrumental flux and
    flux error to nanoJansky.

Returns
-------
dcrModel : `lsst.pipe.tasks.DcrModel`
    Best fit model of the true sky after correcting chromatic effects.

Definition at line 65 of file dcrModel.py.

                  filterLabel=None, psf=None, photoCalib=None):
        """Initialize a DcrModel by dividing a coadd between the subfilters.
 
        Parameters
        ----------
        maskedImage : `lsst.afw.image.MaskedImage`
            Input coadded image to divide equally between the subfilters.
        dcrNumSubfilters : `int`
            Number of sub-filters used to model chromatic effects within a
            band.
        effectiveWavelength : `float`
            The effective wavelengths of the current filter, in nanometers.
        bandwidth : `float`
            The bandwidth of the current filter, in nanometers.
        filterLabel : `lsst.afw.image.FilterLabel`, optional
            The filter label, set in the current instruments' obs package.
            Required for any calculation of DCR, including making matched
            templates.
        psf : `lsst.afw.detection.Psf`, optional
            Point spread function (PSF) of the model.
            Required if the ``DcrModel`` will be persisted.
        photoCalib : `lsst.afw.image.PhotoCalib`, optional
            Calibration to convert instrumental flux and
            flux error to nanoJansky.
 
        Returns
        -------
        dcrModel : `lsst.pipe.tasks.DcrModel`
            Best fit model of the true sky after correcting chromatic effects.
        """
        # NANs will potentially contaminate the entire image,
        # depending on the shift or convolution type used.
        model = maskedImage.image.clone()
        mask = maskedImage.mask.clone()
        # We divide the variance by N and not N**2 because we will assume each
        # subfilter is independent. That means that the significance of
        # detected sources will be lower by a factor of sqrt(N) in the
        # subfilter images, but we will recover it when we combine the
        # subfilter images to construct matched templates.
        variance = maskedImage.variance.clone()
        variance /= dcrNumSubfilters
        model /= dcrNumSubfilters
        modelImages = [model, ]
        for subfilter in range(1, dcrNumSubfilters):
            modelImages.append(model.clone())
        return cls(modelImages, effectiveWavelength, bandwidth,
                   filterLabel=filterLabel, psf=psf, mask=mask, variance=variance, photoCalib=photoCalib)
 

fromQuantum()

def lsst.ip.diffim.dcrModel.DcrModel.fromQuantum	(		cls,
			availableCoaddRefs,
			effectiveWavelength,
			bandwidth
	)

Load an existing DcrModel from a Gen 3 repository.

Parameters
----------
availableCoaddRefs : `dict` [`int`, `lsst.daf.butler.DeferredDatasetHandle`]
    Dictionary of spatially relevant retrieved coadd patches,
    indexed by their sequential patch number.
effectiveWavelength : `float`
    The effective wavelengths of the current filter, in nanometers.
bandwidth : `float`
    The bandwidth of the current filter, in nanometers.

Returns
-------
dcrModel : `lsst.pipe.tasks.DcrModel`
    Best fit model of the true sky after correcting chromatic effects.

Definition at line 169 of file dcrModel.py.

    def fromQuantum(cls, availableCoaddRefs, effectiveWavelength, bandwidth):
        """Load an existing DcrModel from a Gen 3 repository.
 
        Parameters
        ----------
        availableCoaddRefs : `dict` [`int`, `lsst.daf.butler.DeferredDatasetHandle`]
            Dictionary of spatially relevant retrieved coadd patches,
            indexed by their sequential patch number.
        effectiveWavelength : `float`
            The effective wavelengths of the current filter, in nanometers.
        bandwidth : `float`
            The bandwidth of the current filter, in nanometers.
 
        Returns
        -------
        dcrModel : `lsst.pipe.tasks.DcrModel`
            Best fit model of the true sky after correcting chromatic effects.
        """
        filterLabel = None
        psf = None
        mask = None
        variance = None
        photoCalib = None
        modelImages = [None]*len(availableCoaddRefs)
 
        for coaddRef in availableCoaddRefs:
            subfilter = coaddRef.dataId["subfilter"]
            dcrCoadd = coaddRef.get()
            if filterLabel is None:
                filterLabel = dcrCoadd.getFilterLabel()
            if psf is None:
                psf = dcrCoadd.getPsf()
            if mask is None:
                mask = dcrCoadd.mask
            if variance is None:
                variance = dcrCoadd.variance
            if photoCalib is None:
                photoCalib = dcrCoadd.getPhotoCalib()
            modelImages[subfilter] = dcrCoadd.image
        return cls(modelImages, effectiveWavelength, bandwidth, filterLabel, psf, mask, variance, photoCalib)
 

getReferenceImage()

def lsst.ip.diffim.dcrModel.DcrModel.getReferenceImage	(		self,
			bbox = None
	)

Calculate a reference image from the average of the subfilter
images.

Parameters
----------
bbox : `lsst.afw.geom.Box2I`, optional
    Sub-region of the coadd. Returns the entire image if `None`.

Returns
-------
refImage : `numpy.ndarray`
    The reference image with no chromatic effects applied.

Definition at line 346 of file dcrModel.py.

    def getReferenceImage(self, bbox=None):
        """Calculate a reference image from the average of the subfilter
        images.
 
        Parameters
        ----------
        bbox : `lsst.afw.geom.Box2I`, optional
            Sub-region of the coadd. Returns the entire image if `None`.
 
        Returns
        -------
        refImage : `numpy.ndarray`
            The reference image with no chromatic effects applied.
        """
        bbox = bbox or self.bbox
        return np.mean([model[bbox].array for model in self], axis=0)
 

mask()

def lsst.ip.diffim.dcrModel.DcrModel.mask

(

self

)

Return the common mask of each subfilter image.

Returns
-------
mask : `lsst.afw.image.Mask`
    Mask plane of the DCR model.

Definition at line 325 of file dcrModel.py.

    def mask(self):
        """Return the common mask of each subfilter image.
 
        Returns
        -------
        mask : `lsst.afw.image.Mask`
            Mask plane of the DCR model.
        """
        return self._mask
 

psf()

def lsst.ip.diffim.dcrModel.DcrModel.psf

(

self

)

Return the psf of the model.

Returns
-------
psf : `lsst.afw.detection.Psf`
    Point spread function (PSF) of the model.

Definition at line 303 of file dcrModel.py.

    def psf(self):
        """Return the psf of the model.
 
        Returns
        -------
        psf : `lsst.afw.detection.Psf`
            Point spread function (PSF) of the model.
        """
        return self._psf
 

regularizeModelFreq()

def lsst.ip.diffim.dcrModel.DcrModel.regularizeModelFreq	(		self,
			modelImages,
			bbox,
			statsCtrl,
			regularizationFactor,
			regularizationWidth = 2,
			mask = None,
			convergenceMaskPlanes = "DETECTED"
	)

Restrict large variations in the model between subfilters.

Parameters
----------
modelImages : `list` of `lsst.afw.image.Image`
    The new DCR model images from the current iteration.
    The values will be modified in place.
bbox : `lsst.afw.geom.Box2I`
    Sub-region to coadd
statsCtrl : `lsst.afw.math.StatisticsControl`
    Statistics control object for coaddition.
regularizationFactor : `float`
    Maximum relative change of the model allowed between subfilters.
regularizationWidth : `int`, optional
    Minimum radius of a region to include in regularization, in pixels.
mask : `lsst.afw.image.Mask`, optional
    Optional alternate mask
convergenceMaskPlanes : `list` of `str`, or `str`, optional
    Mask planes to use to calculate convergence.

Notes
-----
This implementation of frequency regularization restricts each
subfilter image to be a smoothly-varying function times a reference
image.

Definition at line 547 of file dcrModel.py.

                            regularizationWidth=2, mask=None, convergenceMaskPlanes="DETECTED"):
        """Restrict large variations in the model between subfilters.
 
        Parameters
        ----------
        modelImages : `list` of `lsst.afw.image.Image`
            The new DCR model images from the current iteration.
            The values will be modified in place.
        bbox : `lsst.afw.geom.Box2I`
            Sub-region to coadd
        statsCtrl : `lsst.afw.math.StatisticsControl`
            Statistics control object for coaddition.
        regularizationFactor : `float`
            Maximum relative change of the model allowed between subfilters.
        regularizationWidth : `int`, optional
            Minimum radius of a region to include in regularization, in pixels.
        mask : `lsst.afw.image.Mask`, optional
            Optional alternate mask
        convergenceMaskPlanes : `list` of `str`, or `str`, optional
            Mask planes to use to calculate convergence.
 
        Notes
        -----
        This implementation of frequency regularization restricts each
        subfilter image to be a smoothly-varying function times a reference
        image.
        """
        # ``regularizationFactor`` is the maximum change between subfilter
        # images, so the maximum difference between one subfilter image and the
        # average will be the square root of that.
        maxDiff = np.sqrt(regularizationFactor)
        noiseLevel = self.calculateNoiseCutoff(modelImages[0], statsCtrl, bufferSize=5, mask=mask, bbox=bbox)
        referenceImage = self.getReferenceImage(bbox)
        badPixels = np.isnan(referenceImage) | (referenceImage <= 0.)
        if np.sum(~badPixels) == 0:
            # Skip regularization if there are no valid pixels
            return
        referenceImage[badPixels] = 0.
        filterWidth = regularizationWidth
        fwhm = 2.*filterWidth
        # The noise should be lower in the smoothed image by
        # sqrt(Nsmooth) ~ fwhm pixels
        noiseLevel /= fwhm
        smoothRef = ndimage.filters.gaussian_filter(referenceImage, filterWidth, mode='constant')
        # Add a three sigma offset to both the reference and model to prevent
        # dividing by zero. Note that this will also slightly suppress faint
        # variations in color.
        smoothRef += 3.*noiseLevel
 
        lowThreshold = smoothRef/maxDiff
        highThreshold = smoothRef*maxDiff
        for model in modelImages:
            self.applyImageThresholds(model.array,
                                      highThreshold=highThreshold,
                                      lowThreshold=lowThreshold,
                                      regularizationWidth=regularizationWidth)
            smoothModel = ndimage.filters.gaussian_filter(model.array, filterWidth, mode='constant')
            smoothModel += 3.*noiseLevel
            relativeModel = smoothModel/smoothRef
            # Now sharpen the smoothed relativeModel using an alpha of 3.
            alpha = 3.
            relativeModel2 = ndimage.filters.gaussian_filter(relativeModel, filterWidth/alpha)
            relativeModel += alpha*(relativeModel - relativeModel2)
            model.array = relativeModel*referenceImage
 

regularizeModelIter()

def lsst.ip.diffim.dcrModel.DcrModel.regularizeModelIter	(		self,
			subfilter,
			newModel,
			bbox,
			regularizationFactor,
			regularizationWidth = 2
	)

Restrict large variations in the model between iterations.

Parameters
----------
subfilter : `int`
    Index of the current subfilter within the full band.
newModel : `lsst.afw.image.Image`
    The new DCR model for one subfilter from the current iteration.
    Values in ``newModel`` that are extreme compared with the last
    iteration are modified in place.
bbox : `lsst.afw.geom.Box2I`
    Sub-region to coadd
regularizationFactor : `float`
    Maximum relative change of the model allowed between iterations.
regularizationWidth : int, optional
    Minimum radius of a region to include in regularization, in pixels.

Definition at line 521 of file dcrModel.py.

                            regularizationWidth=2):
        """Restrict large variations in the model between iterations.
 
        Parameters
        ----------
        subfilter : `int`
            Index of the current subfilter within the full band.
        newModel : `lsst.afw.image.Image`
            The new DCR model for one subfilter from the current iteration.
            Values in ``newModel`` that are extreme compared with the last
            iteration are modified in place.
        bbox : `lsst.afw.geom.Box2I`
            Sub-region to coadd
        regularizationFactor : `float`
            Maximum relative change of the model allowed between iterations.
        regularizationWidth : int, optional
            Minimum radius of a region to include in regularization, in pixels.
        """
        refImage = self[subfilter][bbox].array
        highThreshold = np.abs(refImage)*regularizationFactor
        lowThreshold = refImage/regularizationFactor
        newImage = newModel.array
        self.applyImageThresholds(newImage, highThreshold=highThreshold, lowThreshold=lowThreshold,
                                  regularizationWidth=regularizationWidth)
 

variance()

def lsst.ip.diffim.dcrModel.DcrModel.variance

(

self

)

Return the common variance of each subfilter image.

Returns
-------
variance : `lsst.afw.image.Image`
    Variance plane of the DCR model.

Definition at line 336 of file dcrModel.py.

    def variance(self):
        """Return the common variance of each subfilter image.
 
        Returns
        -------
        variance : `lsst.afw.image.Image`
            Variance plane of the DCR model.
        """
        return self._variance
 

Member Data Documentation

dcrNumSubfilters

lsst.ip.diffim.dcrModel.DcrModel.dcrNumSubfilters

Definition at line 54 of file dcrModel.py.

modelImages

lsst.ip.diffim.dcrModel.DcrModel.modelImages

Definition at line 55 of file dcrModel.py.

photoCalib

lsst.ip.diffim.dcrModel.DcrModel.photoCalib

Definition at line 62 of file dcrModel.py.

---

The documentation for this class was generated from the following file:

• /j/snowflake/release/lsstsw/stack/lsst-scipipe-0.7.0/Linux64/ip_diffim/22.0.1-24-g1ad7a390+a9625a72a8/python/lsst/ip/diffim/dcrModel.py