LSSTApplications  17.0+124,17.0+14,17.0+73,18.0.0+37,18.0.0+80,18.0.0-4-g68ffd23+4,18.1.0-1-g0001055+12,18.1.0-1-g03d53ef+5,18.1.0-1-g1349e88+55,18.1.0-1-g2505f39+44,18.1.0-1-g5315e5e+4,18.1.0-1-g5e4b7ea+14,18.1.0-1-g7e8fceb+4,18.1.0-1-g85f8cd4+48,18.1.0-1-g8ff0b9f+4,18.1.0-1-ga2c679d+1,18.1.0-1-gd55f500+35,18.1.0-10-gb58edde+2,18.1.0-11-g0997b02+4,18.1.0-13-gfe4edf0b+12,18.1.0-14-g259bd21+21,18.1.0-19-gdb69f3f+2,18.1.0-2-g5f9922c+24,18.1.0-2-gd3b74e5+11,18.1.0-2-gfbf3545+32,18.1.0-26-g728bddb4+5,18.1.0-27-g6ff7ca9+2,18.1.0-3-g52aa583+25,18.1.0-3-g8ea57af+9,18.1.0-3-gb69f684+42,18.1.0-3-gfcaddf3+6,18.1.0-32-gd8786685a,18.1.0-4-gf3f9b77+6,18.1.0-5-g1dd662b+2,18.1.0-5-g6dbcb01+41,18.1.0-6-gae77429+3,18.1.0-7-g9d75d83+9,18.1.0-7-gae09a6d+30,18.1.0-9-gc381ef5+4,w.2019.45
LSSTDataManagementBasePackage
dcrModel.py
Go to the documentation of this file.
1 # This file is part of ip_diffim.
2 #
3 # LSST Data Management System
4 # This product includes software developed by the
5 # LSST Project (http://www.lsst.org/).
6 # See COPYRIGHT file at the top of the source tree.
7 #
8 # This program is free software: you can redistribute it and/or modify
9 # it under the terms of the GNU General Public License as published by
10 # the Free Software Foundation, either version 3 of the License, or
11 # (at your option) any later version.
12 #
13 # This program is distributed in the hope that it will be useful,
14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 # GNU General Public License for more details.
17 #
18 # You should have received a copy of the LSST License Statement and
19 # the GNU General Public License along with this program. If not,
20 # see <https://www.lsstcorp.org/LegalNotices/>.
21 #
22 
23 import numpy as np
24 from scipy import ndimage
25 from lsst.afw.coord.refraction import differentialRefraction
26 import lsst.afw.image as afwImage
27 import lsst.geom as geom
28 
29 __all__ = ["DcrModel", "applyDcr", "calculateDcr", "calculateImageParallacticAngle"]
30 
31 
32 class DcrModel:
33  """A model of the true sky after correcting chromatic effects.
34 
35  Attributes
36  ----------
37  dcrNumSubfilters : `int`
38  Number of sub-filters used to model chromatic effects within a band.
39  modelImages : `list` of `lsst.afw.image.Image`
40  A list of masked images, each containing the model for one subfilter
41 
42  Notes
43  -----
44  The ``DcrModel`` contains an estimate of the true sky, at a higher
45  wavelength resolution than the input observations. It can be forward-
46  modeled to produce Differential Chromatic Refraction (DCR) matched
47  templates for a given ``Exposure``, and provides utilities for conditioning
48  the model in ``dcrAssembleCoadd`` to avoid oscillating solutions between
49  iterations of forward modeling or between the subfilters of the model.
50  """
51 
52  def __init__(self, modelImages, filterInfo=None, psf=None, mask=None, variance=None):
53  self.dcrNumSubfilters = len(modelImages)
54  self.modelImages = modelImages
55  self._filter = filterInfo
56  self._psf = psf
57  self._mask = mask
58  self._variance = variance
59 
60  @classmethod
61  def fromImage(cls, maskedImage, dcrNumSubfilters, filterInfo=None, psf=None):
62  """Initialize a DcrModel by dividing a coadd between the subfilters.
63 
64  Parameters
65  ----------
66  maskedImage : `lsst.afw.image.MaskedImage`
67  Input coadded image to divide equally between the subfilters.
68  dcrNumSubfilters : `int`
69  Number of sub-filters used to model chromatic effects within a band.
70  filterInfo : `lsst.afw.image.Filter`, optional
71  The filter definition, set in the current instruments' obs package.
72  Required for any calculation of DCR, including making matched templates.
73  psf : `lsst.afw.detection.Psf`, optional
74  Point spread function (PSF) of the model.
75  Required if the ``DcrModel`` will be persisted.
76 
77  Returns
78  -------
79  dcrModel : `lsst.pipe.tasks.DcrModel`
80  Best fit model of the true sky after correcting chromatic effects.
81 
82  Raises
83  ------
84  ValueError
85  If there are any unmasked NAN values in ``maskedImage``.
86  """
87  # NANs will potentially contaminate the entire image,
88  # depending on the shift or convolution type used.
89  model = maskedImage.image.clone()
90  mask = maskedImage.mask.clone()
91  # We divide the variance by N and not N**2 because we will assume each
92  # subfilter is independent. That means that the significance of
93  # detected sources will be lower by a factor of sqrt(N) in the
94  # subfilter images, but we will recover it when we combine the
95  # subfilter images to construct matched templates.
96  variance = maskedImage.variance.clone()
97  variance /= dcrNumSubfilters
98  model /= dcrNumSubfilters
99  modelImages = [model, ]
100  for subfilter in range(1, dcrNumSubfilters):
101  modelImages.append(model.clone())
102  return cls(modelImages, filterInfo, psf, mask, variance)
103 
104  @classmethod
105  def fromDataRef(cls, dataRef, datasetType="dcrCoadd", numSubfilters=None, **kwargs):
106  """Load an existing DcrModel from a repository.
107 
108  Parameters
109  ----------
110  dataRef : `lsst.daf.persistence.ButlerDataRef`
111  Data reference defining the patch for coaddition and the
112  reference Warp
113  datasetType : `str`, optional
114  Name of the DcrModel in the registry {"dcrCoadd", "dcrCoadd_sub"}
115  numSubfilters : `int`
116  Number of sub-filters used to model chromatic effects within a band.
117  **kwargs
118  Additional keyword arguments to pass to look up the model in the data registry.
119  Common keywords and their types include: ``tract``:`str`, ``patch``:`str`,
120  ``bbox``:`lsst.afw.geom.Box2I`
121 
122  Returns
123  -------
124  dcrModel : `lsst.pipe.tasks.DcrModel`
125  Best fit model of the true sky after correcting chromatic effects.
126  """
127  modelImages = []
128  filterInfo = None
129  psf = None
130  mask = None
131  variance = None
132  for subfilter in range(numSubfilters):
133  dcrCoadd = dataRef.get(datasetType, subfilter=subfilter,
134  numSubfilters=numSubfilters, **kwargs)
135  if filterInfo is None:
136  filterInfo = dcrCoadd.getFilter()
137  if psf is None:
138  psf = dcrCoadd.getPsf()
139  if mask is None:
140  mask = dcrCoadd.mask
141  if variance is None:
142  variance = dcrCoadd.variance
143  modelImages.append(dcrCoadd.image)
144  return cls(modelImages, filterInfo, psf, mask, variance)
145 
146  def __len__(self):
147  """Return the number of subfilters.
148 
149  Returns
150  -------
151  dcrNumSubfilters : `int`
152  The number of DCR subfilters in the model.
153  """
154  return self.dcrNumSubfilters
155 
156  def __getitem__(self, subfilter):
157  """Iterate over the subfilters of the DCR model.
158 
159  Parameters
160  ----------
161  subfilter : `int`
162  Index of the current ``subfilter`` within the full band.
163  Negative indices are allowed, and count in reverse order
164  from the highest ``subfilter``.
165 
166  Returns
167  -------
168  modelImage : `lsst.afw.image.Image`
169  The DCR model for the given ``subfilter``.
170 
171  Raises
172  ------
173  IndexError
174  If the requested ``subfilter`` is greater or equal to the number
175  of subfilters in the model.
176  """
177  if np.abs(subfilter) >= len(self):
178  raise IndexError("subfilter out of bounds.")
179  return self.modelImages[subfilter]
180 
181  def __setitem__(self, subfilter, maskedImage):
182  """Update the model image for one subfilter.
183 
184  Parameters
185  ----------
186  subfilter : `int`
187  Index of the current subfilter within the full band.
188  maskedImage : `lsst.afw.image.Image`
189  The DCR model to set for the given ``subfilter``.
190 
191  Raises
192  ------
193  IndexError
194  If the requested ``subfilter`` is greater or equal to the number
195  of subfilters in the model.
196  ValueError
197  If the bounding box of the new image does not match.
198  """
199  if np.abs(subfilter) >= len(self):
200  raise IndexError("subfilter out of bounds.")
201  if maskedImage.getBBox() != self.bbox:
202  raise ValueError("The bounding box of a subfilter must not change.")
203  self.modelImages[subfilter] = maskedImage
204 
205  @property
206  def filter(self):
207  """Return the filter of the model.
208 
209  Returns
210  -------
211  filter : `lsst.afw.image.Filter`
212  The filter definition, set in the current instruments' obs package.
213  """
214  return self._filter
215 
216  @property
217  def psf(self):
218  """Return the psf of the model.
219 
220  Returns
221  -------
222  psf : `lsst.afw.detection.Psf`
223  Point spread function (PSF) of the model.
224  """
225  return self._psf
226 
227  @property
228  def bbox(self):
229  """Return the common bounding box of each subfilter image.
230 
231  Returns
232  -------
233  bbox : `lsst.afw.geom.Box2I`
234  Bounding box of the DCR model.
235  """
236  return self[0].getBBox()
237 
238  @property
239  def mask(self):
240  """Return the common mask of each subfilter image.
241 
242  Returns
243  -------
244  mask : `lsst.afw.image.Mask`
245  Mask plane of the DCR model.
246  """
247  return self._mask
248 
249  @property
250  def variance(self):
251  """Return the common variance of each subfilter image.
252 
253  Returns
254  -------
255  variance : `lsst.afw.image.Image`
256  Variance plane of the DCR model.
257  """
258  return self._variance
259 
260  def getReferenceImage(self, bbox=None):
261  """Calculate a reference image from the average of the subfilter images.
262 
263  Parameters
264  ----------
265  bbox : `lsst.afw.geom.Box2I`, optional
266  Sub-region of the coadd. Returns the entire image if `None`.
267 
268  Returns
269  -------
270  refImage : `numpy.ndarray`
271  The reference image with no chromatic effects applied.
272  """
273  bbox = bbox or self.bbox
274  return np.mean([model[bbox].array for model in self], axis=0)
275 
276  def assign(self, dcrSubModel, bbox=None):
277  """Update a sub-region of the ``DcrModel`` with new values.
278 
279  Parameters
280  ----------
281  dcrSubModel : `lsst.pipe.tasks.DcrModel`
282  New model of the true scene after correcting chromatic effects.
283  bbox : `lsst.afw.geom.Box2I`, optional
284  Sub-region of the coadd.
285  Defaults to the bounding box of ``dcrSubModel``.
286 
287  Raises
288  ------
289  ValueError
290  If the new model has a different number of subfilters.
291  """
292  if len(dcrSubModel) != len(self):
293  raise ValueError("The number of DCR subfilters must be the same "
294  "between the old and new models.")
295  bbox = bbox or self.bbox
296  for model, subModel in zip(self, dcrSubModel):
297  model.assign(subModel[bbox], bbox)
298 
299  def buildMatchedTemplate(self, exposure=None, order=3,
300  visitInfo=None, bbox=None, wcs=None, mask=None,
301  splitSubfilters=True, splitThreshold=0., amplifyModel=1.):
302  """Create a DCR-matched template image for an exposure.
303 
304  Parameters
305  ----------
306  exposure : `lsst.afw.image.Exposure`, optional
307  The input exposure to build a matched template for.
308  May be omitted if all of the metadata is supplied separately
309  order : `int`, optional
310  Interpolation order of the DCR shift.
311  visitInfo : `lsst.afw.image.VisitInfo`, optional
312  Metadata for the exposure. Ignored if ``exposure`` is set.
313  bbox : `lsst.afw.geom.Box2I`, optional
314  Sub-region of the coadd. Ignored if ``exposure`` is set.
315  wcs : `lsst.afw.geom.SkyWcs`, optional
316  Coordinate system definition (wcs) for the exposure.
317  Ignored if ``exposure`` is set.
318  mask : `lsst.afw.image.Mask`, optional
319  reference mask to use for the template image.
320  splitSubfilters : `bool`, optional
321  Calculate DCR for two evenly-spaced wavelengths in each subfilter,
322  instead of at the midpoint. Default: True
323  splitThreshold : `float`, optional
324  Minimum DCR difference within a subfilter required to use ``splitSubfilters``
325  amplifyModel : `float`, optional
326  Multiplication factor to amplify differences between model planes.
327  Used to speed convergence of iterative forward modeling.
328 
329  Returns
330  -------
331  templateImage : `lsst.afw.image.ImageF`
332  The DCR-matched template
333 
334  Raises
335  ------
336  ValueError
337  If neither ``exposure`` or all of ``visitInfo``, ``bbox``, and ``wcs`` are set.
338  """
339  if self.filter is None:
340  raise ValueError("'filterInfo' must be set for the DcrModel in order to calculate DCR.")
341  if exposure is not None:
342  visitInfo = exposure.getInfo().getVisitInfo()
343  bbox = exposure.getBBox()
344  wcs = exposure.getInfo().getWcs()
345  elif visitInfo is None or bbox is None or wcs is None:
346  raise ValueError("Either exposure or visitInfo, bbox, and wcs must be set.")
347  dcrShift = calculateDcr(visitInfo, wcs, self.filter, len(self), splitSubfilters=splitSubfilters)
348  templateImage = afwImage.ImageF(bbox)
349  refModel = self.getReferenceImage(bbox)
350  for subfilter, dcr in enumerate(dcrShift):
351  if amplifyModel > 1:
352  model = (self[subfilter][bbox].array - refModel)*amplifyModel + refModel
353  else:
354  model = self[subfilter][bbox].array
355  templateImage.array += applyDcr(model, dcr, splitSubfilters=splitSubfilters,
356  splitThreshold=splitThreshold, order=order)
357  return templateImage
358 
359  def buildMatchedExposure(self, exposure=None,
360  visitInfo=None, bbox=None, wcs=None, mask=None):
361  """Wrapper to create an exposure from a template image.
362 
363  Parameters
364  ----------
365  exposure : `lsst.afw.image.Exposure`, optional
366  The input exposure to build a matched template for.
367  May be omitted if all of the metadata is supplied separately
368  visitInfo : `lsst.afw.image.VisitInfo`, optional
369  Metadata for the exposure. Ignored if ``exposure`` is set.
370  bbox : `lsst.afw.geom.Box2I`, optional
371  Sub-region of the coadd. Ignored if ``exposure`` is set.
372  wcs : `lsst.afw.geom.SkyWcs`, optional
373  Coordinate system definition (wcs) for the exposure.
374  Ignored if ``exposure`` is set.
375  mask : `lsst.afw.image.Mask`, optional
376  reference mask to use for the template image.
377 
378  Returns
379  -------
380  templateExposure : `lsst.afw.image.exposureF`
381  The DCR-matched template
382  """
383  if bbox is None:
384  bbox = exposure.getBBox()
385  templateImage = self.buildMatchedTemplate(exposure=exposure, visitInfo=visitInfo,
386  bbox=bbox, wcs=wcs, mask=mask)
387  maskedImage = afwImage.MaskedImageF(bbox)
388  maskedImage.image = templateImage[bbox]
389  maskedImage.mask = self.mask[bbox]
390  maskedImage.variance = self.variance[bbox]
391  templateExposure = afwImage.ExposureF(bbox, wcs)
392  templateExposure.setMaskedImage(maskedImage[bbox])
393  templateExposure.setPsf(self.psf)
394  templateExposure.setFilter(self.filter)
395  return templateExposure
396 
397  def conditionDcrModel(self, modelImages, bbox, gain=1.):
398  """Average two iterations' solutions to reduce oscillations.
399 
400  Parameters
401  ----------
402  modelImages : `list` of `lsst.afw.image.Image`
403  The new DCR model images from the current iteration.
404  The values will be modified in place.
405  bbox : `lsst.afw.geom.Box2I`
406  Sub-region of the coadd
407  gain : `float`, optional
408  Relative weight to give the new solution when updating the model.
409  Defaults to 1.0, which gives equal weight to both solutions.
410  """
411  # Calculate weighted averages of the images.
412  for model, newModel in zip(self, modelImages):
413  newModel *= gain
414  newModel += model[bbox]
415  newModel /= 1. + gain
416 
417  def regularizeModelIter(self, subfilter, newModel, bbox, regularizationFactor,
418  regularizationWidth=2):
419  """Restrict large variations in the model between iterations.
420 
421  Parameters
422  ----------
423  subfilter : `int`
424  Index of the current subfilter within the full band.
425  newModel : `lsst.afw.image.Image`
426  The new DCR model for one subfilter from the current iteration.
427  Values in ``newModel`` that are extreme compared with the last
428  iteration are modified in place.
429  bbox : `lsst.afw.geom.Box2I`
430  Sub-region to coadd
431  regularizationFactor : `float`
432  Maximum relative change of the model allowed between iterations.
433  regularizationWidth : int, optional
434  Minimum radius of a region to include in regularization, in pixels.
435  """
436  refImage = self[subfilter][bbox].array
437  highThreshold = np.abs(refImage)*regularizationFactor
438  lowThreshold = refImage/regularizationFactor
439  newImage = newModel.array
440  self.applyImageThresholds(newImage, highThreshold=highThreshold, lowThreshold=lowThreshold,
441  regularizationWidth=regularizationWidth)
442 
443  def regularizeModelFreq(self, modelImages, bbox, statsCtrl, regularizationFactor,
444  regularizationWidth=2, mask=None, convergenceMaskPlanes="DETECTED"):
445  """Restrict large variations in the model between subfilters.
446 
447  Parameters
448  ----------
449  modelImages : `list` of `lsst.afw.image.Image`
450  The new DCR model images from the current iteration.
451  The values will be modified in place.
452  bbox : `lsst.afw.geom.Box2I`
453  Sub-region to coadd
454  statsCtrl : `lsst.afw.math.StatisticsControl`
455  Statistics control object for coaddition.
456  regularizationFactor : `float`
457  Maximum relative change of the model allowed between subfilters.
458  regularizationWidth : `int`, optional
459  Minimum radius of a region to include in regularization, in pixels.
460  mask : `lsst.afw.image.Mask`, optional
461  Optional alternate mask
462  convergenceMaskPlanes : `list` of `str`, or `str`, optional
463  Mask planes to use to calculate convergence.
464 
465  Notes
466  -----
467  This implementation of frequency regularization restricts each subfilter
468  image to be a smoothly-varying function times a reference image.
469  """
470  # ``regularizationFactor`` is the maximum change between subfilter images, so the maximum difference
471  # between one subfilter image and the average will be the square root of that.
472  maxDiff = np.sqrt(regularizationFactor)
473  noiseLevel = self.calculateNoiseCutoff(modelImages[0], statsCtrl, bufferSize=5, mask=mask, bbox=bbox)
474  referenceImage = self.getReferenceImage(bbox)
475  badPixels = np.isnan(referenceImage) | (referenceImage <= 0.)
476  if np.sum(~badPixels) == 0:
477  # Skip regularization if there are no valid pixels
478  return
479  referenceImage[badPixels] = 0.
480  filterWidth = regularizationWidth
481  fwhm = 2.*filterWidth
482  # The noise should be lower in the smoothed image by sqrt(Nsmooth) ~ fwhm pixels
483  noiseLevel /= fwhm
484  smoothRef = ndimage.filters.gaussian_filter(referenceImage, filterWidth, mode='constant')
485  # Add a three sigma offset to both the reference and model to prevent dividing by zero.
486  # Note that this will also slightly suppress faint variations in color.
487  smoothRef += 3.*noiseLevel
488 
489  lowThreshold = smoothRef/maxDiff
490  highThreshold = smoothRef*maxDiff
491  for model in modelImages:
492  self.applyImageThresholds(model.array,
493  highThreshold=highThreshold,
494  lowThreshold=lowThreshold,
495  regularizationWidth=regularizationWidth)
496  smoothModel = ndimage.filters.gaussian_filter(model.array, filterWidth, mode='constant')
497  smoothModel += 3.*noiseLevel
498  relativeModel = smoothModel/smoothRef
499  # Now sharpen the smoothed relativeModel using an alpha of 3.
500  alpha = 3.
501  relativeModel2 = ndimage.filters.gaussian_filter(relativeModel, filterWidth/alpha)
502  relativeModel += alpha*(relativeModel - relativeModel2)
503  model.array = relativeModel*referenceImage
504 
505  def calculateNoiseCutoff(self, image, statsCtrl, bufferSize,
506  convergenceMaskPlanes="DETECTED", mask=None, bbox=None):
507  """Helper function to calculate the background noise level of an image.
508 
509  Parameters
510  ----------
511  image : `lsst.afw.image.Image`
512  The input image to evaluate the background noise properties.
513  statsCtrl : `lsst.afw.math.StatisticsControl`
514  Statistics control object for coaddition.
515  bufferSize : `int`
516  Number of additional pixels to exclude
517  from the edges of the bounding box.
518  convergenceMaskPlanes : `list` of `str`, or `str`
519  Mask planes to use to calculate convergence.
520  mask : `lsst.afw.image.Mask`, Optional
521  Optional alternate mask
522  bbox : `lsst.afw.geom.Box2I`, optional
523  Sub-region of the masked image to calculate the noise level over.
524 
525  Returns
526  -------
527  noiseCutoff : `float`
528  The threshold value to treat pixels as noise in an image..
529  """
530  if bbox is None:
531  bbox = self.bbox
532  if mask is None:
533  mask = self.mask[bbox]
534  bboxShrink = geom.Box2I(bbox)
535  bboxShrink.grow(-bufferSize)
536  convergeMask = mask.getPlaneBitMask(convergenceMaskPlanes)
537 
538  backgroundPixels = mask[bboxShrink].array & (statsCtrl.getAndMask() | convergeMask) == 0
539  noiseCutoff = np.std(image[bboxShrink].array[backgroundPixels])
540  return noiseCutoff
541 
542  def applyImageThresholds(self, image, highThreshold=None, lowThreshold=None, regularizationWidth=2):
543  """Restrict image values to be between upper and lower limits.
544 
545  This method flags all pixels in an image that are outside of the given
546  threshold values. The threshold values are taken from a reference image,
547  so noisy pixels are likely to get flagged. In order to exclude those
548  noisy pixels, the array of flags is eroded and dilated, which removes
549  isolated pixels outside of the thresholds from the list of pixels to be
550  modified. Pixels that remain flagged after this operation have their
551  values set to the appropriate upper or lower threshold value.
552 
553  Parameters
554  ----------
555  image : `numpy.ndarray`
556  The image to apply the thresholds to.
557  The values will be modified in place.
558  highThreshold : `numpy.ndarray`, optional
559  Array of upper limit values for each pixel of ``image``.
560  lowThreshold : `numpy.ndarray`, optional
561  Array of lower limit values for each pixel of ``image``.
562  regularizationWidth : `int`, optional
563  Minimum radius of a region to include in regularization, in pixels.
564  """
565  # Generate the structure for binary erosion and dilation, which is used to remove noise-like pixels.
566  # Groups of pixels with a radius smaller than ``regularizationWidth``
567  # will be excluded from regularization.
568  filterStructure = ndimage.iterate_structure(ndimage.generate_binary_structure(2, 1),
569  regularizationWidth)
570  if highThreshold is not None:
571  highPixels = image > highThreshold
572  if regularizationWidth > 0:
573  # Erode and dilate ``highPixels`` to exclude noisy pixels.
574  highPixels = ndimage.morphology.binary_opening(highPixels, structure=filterStructure)
575  image[highPixels] = highThreshold[highPixels]
576  if lowThreshold is not None:
577  lowPixels = image < lowThreshold
578  if regularizationWidth > 0:
579  # Erode and dilate ``lowPixels`` to exclude noisy pixels.
580  lowPixels = ndimage.morphology.binary_opening(lowPixels, structure=filterStructure)
581  image[lowPixels] = lowThreshold[lowPixels]
582 
583 
584 def applyDcr(image, dcr, useInverse=False, splitSubfilters=False, splitThreshold=0.,
585  doPrefilter=True, order=3):
586  """Shift an image along the X and Y directions.
587 
588  Parameters
589  ----------
590  image : `numpy.ndarray`
591  The input image to shift.
592  dcr : `tuple`
593  Shift calculated with ``calculateDcr``.
594  Uses numpy axes ordering (Y, X).
595  If ``splitSubfilters`` is set, each element is itself a `tuple`
596  of two `float`, corresponding to the DCR shift at the two wavelengths.
597  Otherwise, each element is a `float` corresponding to the DCR shift at
598  the effective wavelength of the subfilter.
599  useInverse : `bool`, optional
600  Apply the shift in the opposite direction. Default: False
601  splitSubfilters : `bool`, optional
602  Calculate DCR for two evenly-spaced wavelengths in each subfilter,
603  instead of at the midpoint. Default: False
604  splitThreshold : `float`, optional
605  Minimum DCR difference within a subfilter required to use ``splitSubfilters``
606  doPrefilter : `bool`, optional
607  Spline filter the image before shifting, if set. Filtering is required,
608  so only set to False if the image is already filtered.
609  Filtering takes ~20% of the time of shifting, so if `applyDcr` will be
610  called repeatedly on the same image it is more efficient to precalculate
611  the filter.
612  order : `int`, optional
613  The order of the spline interpolation, default is 3.
614 
615  Returns
616  -------
617  shiftedImage : `numpy.ndarray`
618  A copy of the input image with the specified shift applied.
619  """
620  if doPrefilter:
621  prefilteredImage = ndimage.spline_filter(image, order=order)
622  else:
623  prefilteredImage = image
624  if splitSubfilters:
625  shiftAmp = np.max(np.abs([_dcr0 - _dcr1 for _dcr0, _dcr1 in zip(dcr[0], dcr[1])]))
626  if shiftAmp >= splitThreshold:
627  if useInverse:
628  shift = [-1.*s for s in dcr[0]]
629  shift1 = [-1.*s for s in dcr[1]]
630  else:
631  shift = dcr[0]
632  shift1 = dcr[1]
633  shiftedImage = ndimage.shift(prefilteredImage, shift, prefilter=False, order=order)
634  shiftedImage += ndimage.shift(prefilteredImage, shift1, prefilter=False, order=order)
635  shiftedImage /= 2.
636  return shiftedImage
637  else:
638  # If the difference in the DCR shifts is less than the threshold,
639  # then just use the average shift for efficiency.
640  dcr = (np.mean(dcr[0]), np.mean(dcr[1]))
641  if useInverse:
642  shift = [-1.*s for s in dcr]
643  else:
644  shift = dcr
645  shiftedImage = ndimage.shift(prefilteredImage, shift, prefilter=False, order=order)
646  return shiftedImage
647 
648 
649 def calculateDcr(visitInfo, wcs, filterInfo, dcrNumSubfilters, splitSubfilters=False):
650  """Calculate the shift in pixels of an exposure due to DCR.
651 
652  Parameters
653  ----------
654  visitInfo : `lsst.afw.image.VisitInfo`
655  Metadata for the exposure.
656  wcs : `lsst.afw.geom.SkyWcs`
657  Coordinate system definition (wcs) for the exposure.
658  filterInfo : `lsst.afw.image.Filter`
659  The filter definition, set in the current instruments' obs package.
660  dcrNumSubfilters : `int`
661  Number of sub-filters used to model chromatic effects within a band.
662  splitSubfilters : `bool`, optional
663  Calculate DCR for two evenly-spaced wavelengths in each subfilter,
664  instead of at the midpoint. Default: False
665 
666  Returns
667  -------
668  dcrShift : `tuple` of two `float`
669  The 2D shift due to DCR, in pixels.
670  Uses numpy axes ordering (Y, X).
671  """
672  rotation = calculateImageParallacticAngle(visitInfo, wcs)
673  dcrShift = []
674  weight = [0.75, 0.25]
675  lambdaEff = filterInfo.getFilterProperty().getLambdaEff()
676  for wl0, wl1 in wavelengthGenerator(filterInfo, dcrNumSubfilters):
677  # Note that diffRefractAmp can be negative, since it's relative to the midpoint of the full band
678  diffRefractAmp0 = differentialRefraction(wavelength=wl0, wavelengthRef=lambdaEff,
679  elevation=visitInfo.getBoresightAzAlt().getLatitude(),
680  observatory=visitInfo.getObservatory(),
681  weather=visitInfo.getWeather())
682  diffRefractAmp1 = differentialRefraction(wavelength=wl1, wavelengthRef=lambdaEff,
683  elevation=visitInfo.getBoresightAzAlt().getLatitude(),
684  observatory=visitInfo.getObservatory(),
685  weather=visitInfo.getWeather())
686  if splitSubfilters:
687  diffRefractPix0 = diffRefractAmp0.asArcseconds()/wcs.getPixelScale().asArcseconds()
688  diffRefractPix1 = diffRefractAmp1.asArcseconds()/wcs.getPixelScale().asArcseconds()
689  diffRefractArr = [diffRefractPix0*weight[0] + diffRefractPix1*weight[1],
690  diffRefractPix0*weight[1] + diffRefractPix1*weight[0]]
691  shiftX = [diffRefractPix*np.sin(rotation.asRadians()) for diffRefractPix in diffRefractArr]
692  shiftY = [diffRefractPix*np.cos(rotation.asRadians()) for diffRefractPix in diffRefractArr]
693  dcrShift.append(((shiftY[0], shiftX[0]), (shiftY[1], shiftX[1])))
694  else:
695  diffRefractAmp = (diffRefractAmp0 + diffRefractAmp1)/2.
696  diffRefractPix = diffRefractAmp.asArcseconds()/wcs.getPixelScale().asArcseconds()
697  shiftX = diffRefractPix*np.sin(rotation.asRadians())
698  shiftY = diffRefractPix*np.cos(rotation.asRadians())
699  dcrShift.append((shiftY, shiftX))
700  return dcrShift
701 
702 
703 def calculateImageParallacticAngle(visitInfo, wcs):
704  """Calculate the total sky rotation angle of an exposure.
705 
706  Parameters
707  ----------
708  visitInfo : `lsst.afw.image.VisitInfo`
709  Metadata for the exposure.
710  wcs : `lsst.afw.geom.SkyWcs`
711  Coordinate system definition (wcs) for the exposure.
712 
713  Returns
714  -------
715  `lsst.geom.Angle`
716  The rotation of the image axis, East from North.
717  Equal to the parallactic angle plus any additional rotation of the
718  coordinate system.
719  A rotation angle of 0 degrees is defined with
720  North along the +y axis and East along the +x axis.
721  A rotation angle of 90 degrees is defined with
722  North along the +x axis and East along the -y axis.
723  """
724  parAngle = visitInfo.getBoresightParAngle().asRadians()
725  cd = wcs.getCdMatrix()
726  if wcs.isFlipped:
727  cdAngle = (np.arctan2(-cd[0, 1], cd[0, 0]) + np.arctan2(cd[1, 0], cd[1, 1]))/2.
728  rotAngle = (cdAngle + parAngle)*geom.radians
729  else:
730  cdAngle = (np.arctan2(cd[0, 1], -cd[0, 0]) + np.arctan2(cd[1, 0], cd[1, 1]))/2.
731  rotAngle = (cdAngle - parAngle)*geom.radians
732  return rotAngle
733 
734 
735 def wavelengthGenerator(filterInfo, dcrNumSubfilters):
736  """Iterate over the wavelength endpoints of subfilters.
737 
738  Parameters
739  ----------
740  filterInfo : `lsst.afw.image.Filter`
741  The filter definition, set in the current instruments' obs package.
742  dcrNumSubfilters : `int`
743  Number of sub-filters used to model chromatic effects within a band.
744 
745  Yields
746  ------
747  `tuple` of two `float`
748  The next set of wavelength endpoints for a subfilter, in nm.
749  """
750  lambdaMin = filterInfo.getFilterProperty().getLambdaMin()
751  lambdaMax = filterInfo.getFilterProperty().getLambdaMax()
752  wlStep = (lambdaMax - lambdaMin)/dcrNumSubfilters
753  for wl in np.linspace(lambdaMin, lambdaMax, dcrNumSubfilters, endpoint=False):
754  yield (wl, wl + wlStep)
def fromImage(cls, maskedImage, dcrNumSubfilters, filterInfo=None, psf=None)
Definition: dcrModel.py:61
def __setitem__(self, subfilter, maskedImage)
Definition: dcrModel.py:181
def calculateImageParallacticAngle(visitInfo, wcs)
Definition: dcrModel.py:703
def applyDcr(image, dcr, useInverse=False, splitSubfilters=False, splitThreshold=0., doPrefilter=True, order=3)
Definition: dcrModel.py:585
def regularizeModelIter(self, subfilter, newModel, bbox, regularizationFactor, regularizationWidth=2)
Definition: dcrModel.py:418
def __getitem__(self, subfilter)
Definition: dcrModel.py:156
def calculateNoiseCutoff(self, image, statsCtrl, bufferSize, convergenceMaskPlanes="DETECTED", mask=None, bbox=None)
Definition: dcrModel.py:506
def __init__(self, modelImages, filterInfo=None, psf=None, mask=None, variance=None)
Definition: dcrModel.py:52
def wavelengthGenerator(filterInfo, dcrNumSubfilters)
Definition: dcrModel.py:735
def applyImageThresholds(self, image, highThreshold=None, lowThreshold=None, regularizationWidth=2)
Definition: dcrModel.py:542
def buildMatchedExposure(self, exposure=None, visitInfo=None, bbox=None, wcs=None, mask=None)
Definition: dcrModel.py:360
def fromDataRef(cls, dataRef, datasetType="dcrCoadd", numSubfilters=None, kwargs)
Definition: dcrModel.py:105
def getReferenceImage(self, bbox=None)
Definition: dcrModel.py:260
def differentialRefraction(wavelength, wavelengthRef, elevation, observatory, weather=None)
Definition: refraction.py:95
def buildMatchedTemplate(self, exposure=None, order=3, visitInfo=None, bbox=None, wcs=None, mask=None, splitSubfilters=True, splitThreshold=0., amplifyModel=1.)
Definition: dcrModel.py:301
def assign(self, dcrSubModel, bbox=None)
Definition: dcrModel.py:276
def calculateDcr(visitInfo, wcs, filterInfo, dcrNumSubfilters, splitSubfilters=False)
Definition: dcrModel.py:649
def regularizeModelFreq(self, modelImages, bbox, statsCtrl, regularizationFactor, regularizationWidth=2, mask=None, convergenceMaskPlanes="DETECTED")
Definition: dcrModel.py:444
def conditionDcrModel(self, modelImages, bbox, gain=1.)
Definition: dcrModel.py:397
Backwards-compatibility support for depersisting the old Calib (FluxMag0/FluxMag0Err) objects...
An integer coordinate rectangle.
Definition: Box.h:55