lsst.meas.algorithms.subtractBackground Namespace Reference

Classes
class	SubtractBackgroundConfig
class	SubtractBackgroundTask
class	TooManyMaskedPixelsError

Functions
	backgroundFlatContext (maskedImage, doApply, backgroundToPhotometricRatio=None)
	filterSuperPixels (bbox, background, superPixelFilterSize=3)

Function Documentation

backgroundFlatContext()

lsst.meas.algorithms.subtractBackground.backgroundFlatContext	(		maskedImage,
			doApply,
			backgroundToPhotometricRatio = None )

Context manager to convert from photometric-flattened to background-
flattened image.

Parameters
----------
maskedImage : `lsst.afw.image.MaskedImage`
    Masked image (image + mask + variance) to convert from a
    photometrically flat image to an image suitable for background
    subtraction.
doApply : `bool`
    Apply the conversion? If False, this context manager will not
    do anything.
backgroundToPhotometricRatio : `lsst.afw.image.Image`, optional
    Image to multiply a photometrically-flattened image by to obtain a
    background-flattened image.
    Only used if ``doApply`` is ``True``.

Yields
------
maskedImage : `lsst.afw.image.MaskedImage`
    Masked image converted into an image suitable for background
    subtraction.

Raises
------
RuntimeError if doApply is True and no ratio is supplied.
ValueError if the ratio is not an `lsst.afw.image.Image`.

Definition at line 43 of file subtractBackground.py.

def backgroundFlatContext(maskedImage, doApply, backgroundToPhotometricRatio=None):
    """Context manager to convert from photometric-flattened to background-
    flattened image.
 
    Parameters
    ----------
    maskedImage : `lsst.afw.image.MaskedImage`
        Masked image (image + mask + variance) to convert from a
        photometrically flat image to an image suitable for background
        subtraction.
    doApply : `bool`
        Apply the conversion? If False, this context manager will not
        do anything.
    backgroundToPhotometricRatio : `lsst.afw.image.Image`, optional
        Image to multiply a photometrically-flattened image by to obtain a
        background-flattened image.
        Only used if ``doApply`` is ``True``.
 
    Yields
    ------
    maskedImage : `lsst.afw.image.MaskedImage`
        Masked image converted into an image suitable for background
        subtraction.
 
    Raises
    ------
    RuntimeError if doApply is True and no ratio is supplied.
    ValueError if the ratio is not an `lsst.afw.image.Image`.
    """
    if doApply:
        if backgroundToPhotometricRatio is None:
            raise RuntimeError("backgroundFlatContext called with doApply=True, "
                               "but without a backgroundToPhotometricRatio")
        if not isinstance(backgroundToPhotometricRatio, afwImage.Image):
            raise ValueError("The backgroundToPhotometricRatio must be an lsst.afw.image.Image")
 
        maskedImage *= backgroundToPhotometricRatio
 
    try:
        yield maskedImage
    finally:
        if doApply:
            maskedImage /= backgroundToPhotometricRatio
 
 

filterSuperPixels()

lsst.meas.algorithms.subtractBackground.filterSuperPixels	(		bbox,
			background,
			superPixelFilterSize = 3 )

Remove outliers from the binned background model.

Parameters
----------
bbox : `lsst.geom.Box2I`
    Bounding box of the original image.
background : `lsst.afw.math.BackgroundMI`
    Fit and binned background image, which will be modified in place.
superPixelFilterSize : `int`, optional
    Size of the median filter to use, in pixels.

Definition at line 416 of file subtractBackground.py.

def filterSuperPixels(bbox, background, superPixelFilterSize=3):
    """Remove outliers from the binned background model.
 
    Parameters
    ----------
    bbox : `lsst.geom.Box2I`
        Bounding box of the original image.
    background : `lsst.afw.math.BackgroundMI`
        Fit and binned background image, which will be modified in place.
    superPixelFilterSize : `int`, optional
        Size of the median filter to use, in pixels.
    """
    statsImg = background.getStatsImage()
    # scipy's median_filter can't handle NaN values
    bad = numpy.isnan(statsImg.image.array)
    if numpy.count_nonzero(bad) > 0:
        statsImg.image.array[bad] = numpy.nanmedian(statsImg.image.array)
    statsImg.image.array = median_filter(statsImg.image.array, mode="reflect", size=superPixelFilterSize)
    background = afwMath.BackgroundMI(bbox, statsImg)