Inheritance diagram for lsst.ip.diffim.makeKernel.MakeKernelTask:
Public Member Functions | |
| __init__ (self, *args, **kwargs) | |
| run (self, template, science, kernelSources, preconvolved=False) | |
| selectKernelSources (self, template, science, candidateList=None, preconvolved=False) | |
| getSelectSources (self, exposure, sigma=None, doSmooth=True, idFactory=None) | |
| makeCandidateList (self, convolved, reference, kernelSize, candidateList, preconvolved=False) | |
| makeKernelBasisList (self, targetFwhmPix=None, referenceFwhmPix=None, basisDegGauss=None, basisSigmaGauss=None, metadata=None) | |
Public Attributes | |
| kConfig | |
| background | |
| selectSchema | |
| selectAlgMetadata | |
Static Public Attributes | |
| ConfigClass = MakeKernelConfig | |
Protected Member Functions | |
| _buildCellSet (self, convolved, reference, candidateList) | |
| _adaptCellSize (self, candidateList) | |
Static Protected Attributes | |
| str | _DefaultName = "makeALKernel" |
Detailed Description
Construct a kernel for PSF matching two exposures.
Definition at line 89 of file makeKernel.py.
Constructor & Destructor Documentation
◆ __init__()
| lsst.ip.diffim.makeKernel.MakeKernelTask.__init__ | ( | self, | |
| * | args, | ||
| ** | kwargs ) |
Create the psf-matching Task
Parameters
----------
*args
Arguments to be passed to ``lsst.pipe.base.task.Task.__init__``
**kwargs
Keyword arguments to be passed to ``lsst.pipe.base.task.Task.__init__``
Notes
-----
The initialization sets the Psf-matching kernel configuration using the value of
self.config.kernel.active. If the kernel is requested with regularization to moderate
the bias/variance tradeoff, currently only used when a delta function kernel basis
is provided, it creates a regularization matrix stored as member variable
self.hMat.
Reimplemented from lsst.ip.diffim.psfMatch.PsfMatchTask.
Definition at line 96 of file makeKernel.py.
96 def __init__(self, *args, **kwargs):
97 PsfMatchTask.__init__(self, *args, **kwargs)
98 self.kConfig = self.config.kernel.active
99 # the background subtraction task uses a config from an unusual location,
100 # so cannot easily be constructed with makeSubtask
101 self.background = SubtractBackgroundTask(config=self.kConfig.afwBackgroundConfig, name="background",
102 parentTask=self)
103 self.selectSchema = lsst.afw.table.SourceTable.makeMinimalSchema()
104 self.selectAlgMetadata = lsst.daf.base.PropertyList()
105 self.makeSubtask("selectDetection", schema=self.selectSchema)
106 self.makeSubtask("selectMeasurement", schema=self.selectSchema, algMetadata=self.selectAlgMetadata)
107
static Schema makeMinimalSchema()
Return a minimal schema for Source tables and records.
Definition Source.h:258
Class for storing ordered metadata with comments.
Definition PropertyList.h:68
Member Function Documentation
◆ _adaptCellSize()
|
protected |
NOT IMPLEMENTED YET.
Parameters
----------
candidateList : `list`
A list of footprints/maskedImages for kernel candidates;
Returns
-------
sizeCellX, sizeCellY : `int`
New dimensions to use for the kernel.
Definition at line 431 of file makeKernel.py.
431 def _adaptCellSize(self, candidateList):
432 """NOT IMPLEMENTED YET.
433
434 Parameters
435 ----------
436 candidateList : `list`
437 A list of footprints/maskedImages for kernel candidates;
438
439 Returns
440 -------
441 sizeCellX, sizeCellY : `int`
442 New dimensions to use for the kernel.
443 """
444 return self.kConfig.sizeCellX, self.kConfig.sizeCellY
◆ _buildCellSet()
|
protected |
Build a SpatialCellSet for use with the solve method.
Parameters
----------
convolved : `lsst.afw.image.MaskedImage`
MaskedImage to PSF-matched to reference.
reference : `lsst.afw.image.MaskedImage`
Reference MaskedImage.
candidateList : `lsst.afw.table.SourceCatalog`
Kernel candidate sources with footprints.
Returns
-------
kernelCellSet : `lsst.afw.math.SpatialCellSet`
A SpatialCellSet for use with self._solve.
Reimplemented from lsst.ip.diffim.psfMatch.PsfMatchTask.
Definition at line 386 of file makeKernel.py.
386 def _buildCellSet(self, convolved, reference, candidateList):
387 """Build a SpatialCellSet for use with the solve method.
388
389 Parameters
390 ----------
391 convolved : `lsst.afw.image.MaskedImage`
392 MaskedImage to PSF-matched to reference.
393 reference : `lsst.afw.image.MaskedImage`
394 Reference MaskedImage.
395 candidateList : `lsst.afw.table.SourceCatalog`
396 Kernel candidate sources with footprints.
397
398 Returns
399 -------
400 kernelCellSet : `lsst.afw.math.SpatialCellSet`
401 A SpatialCellSet for use with self._solve.
402 """
403 sizeCellX, sizeCellY = self._adaptCellSize(candidateList)
404
405 imageBBox = convolved.getBBox()
406 imageBBox.clip(reference.getBBox())
407 # Object to store the KernelCandidates for spatial modeling
408 kernelCellSet = lsst.afw.math.SpatialCellSet(imageBBox, sizeCellX, sizeCellY)
409
410 candidateConfig = lsst.pex.config.makePropertySet(self.kConfig)
411 # Place candidates within the spatial grid
412 for candidate in candidateList:
413 bbox = candidate.getFootprint().getBBox()
414 templateCutout = lsst.afw.image.MaskedImageF(convolved, bbox)
415 scienceCutout = lsst.afw.image.MaskedImageF(reference, bbox)
416
417 kernelCandidate = diffimLib.makeKernelCandidate(candidate,
418 templateCutout,
419 scienceCutout,
420 candidateConfig)
421
422 self.log.debug("Candidate %d at %.2f, %.2f rating=%f",
423 kernelCandidate.getId(),
424 kernelCandidate.getXCenter(),
425 kernelCandidate.getYCenter(),
426 kernelCandidate.getCandidateRating())
427 kernelCellSet.insertCandidate(kernelCandidate)
428
429 return kernelCellSet
430
A collection of SpatialCells covering an entire image.
Definition SpatialCell.h:383
◆ getSelectSources()
| lsst.ip.diffim.makeKernel.MakeKernelTask.getSelectSources | ( | self, | |
| exposure, | |||
| sigma = None, | |||
| doSmooth = True, | |||
| idFactory = None ) |
Get sources to use for Psf-matching.
This method runs detection and measurement on an exposure.
The returned set of sources will be used as candidates for
Psf-matching.
Parameters
----------
exposure : `lsst.afw.image.Exposure`
Exposure on which to run detection/measurement
sigma : `float`, optional
PSF sigma, in pixels, used for smoothing the image for detection.
If `None`, the PSF width will be used.
doSmooth : `bool`
Whether or not to smooth the Exposure with Psf before detection
idFactory : `lsst.afw.table.IdFactory`
Factory for the generation of Source ids
Returns
-------
selectSources :
source catalog containing candidates for the Psf-matching
Definition at line 214 of file makeKernel.py.
214 def getSelectSources(self, exposure, sigma=None, doSmooth=True, idFactory=None):
215 """Get sources to use for Psf-matching.
216
217 This method runs detection and measurement on an exposure.
218 The returned set of sources will be used as candidates for
219 Psf-matching.
220
221 Parameters
222 ----------
223 exposure : `lsst.afw.image.Exposure`
224 Exposure on which to run detection/measurement
225 sigma : `float`, optional
226 PSF sigma, in pixels, used for smoothing the image for detection.
227 If `None`, the PSF width will be used.
228 doSmooth : `bool`
229 Whether or not to smooth the Exposure with Psf before detection
230 idFactory : `lsst.afw.table.IdFactory`
231 Factory for the generation of Source ids
232
233 Returns
234 -------
235 selectSources :
236 source catalog containing candidates for the Psf-matching
237 """
238 if idFactory:
239 table = lsst.afw.table.SourceTable.make(self.selectSchema, idFactory)
240 else:
241 table = lsst.afw.table.SourceTable.make(self.selectSchema)
242 mi = exposure.getMaskedImage()
243
244 imArr = mi.image.array
245 maskArr = mi.mask.array
246 miArr = np.ma.masked_array(imArr, mask=maskArr)
247 try:
248 fitBg = self.background.fitBackground(mi)
249 bkgd = fitBg.getImageF(self.background.config.algorithm,
250 self.background.config.undersampleStyle)
251 except Exception:
252 self.log.warning("Failed to get background model. Falling back to median background estimation")
253 bkgd = np.ma.median(miArr)
254
255 # Take off background for detection
256 mi -= bkgd
257 try:
258 table.setMetadata(self.selectAlgMetadata)
259 detRet = self.selectDetection.run(
260 table=table,
261 exposure=exposure,
262 sigma=sigma,
263 doSmooth=doSmooth
264 )
265 selectSources = detRet.sources
266 self.selectMeasurement.run(measCat=selectSources, exposure=exposure)
267 finally:
268 # Put back on the background in case it is needed down stream
269 mi += bkgd
270 del bkgd
271
272 self.log.info("Selected %d sources via detection measurement.", len(selectSources))
273 return selectSources
274
static std::shared_ptr< SourceTable > make(Schema const &schema, std::shared_ptr< IdFactory > const &idFactory)
Construct a new table.
Definition Source.cc:400
◆ makeCandidateList()
| lsst.ip.diffim.makeKernel.MakeKernelTask.makeCandidateList | ( | self, | |
| convolved, | |||
| reference, | |||
| kernelSize, | |||
| candidateList, | |||
| preconvolved = False ) |
Make a list of acceptable KernelCandidates.
Generate a list of candidate sources for Psf-matching, remove sources
with bad pixel masks set or that extend off the image.
Parameters
----------
convolved : `lsst.afw.image.Exposure`
Exposure that will be convolved. This is typically the template
image, and may have a large bbox than the reference exposure.
reference : `lsst.afw.image.Exposure`
Exposure that will be matched-to. This is typically the science
image.
kernelSize : `float`
Dimensions of the Psf-matching Kernel, used to set detection
footprints.
candidateList : `lsst.afw.table.SourceCatalog`
List of Sources to examine for kernel candidacy.
preconvolved : `bool`, optional
Was the science exposure already convolved with its PSF?
Returns
-------
candidates : `lsst.afw.table.SourceCatalog`
Candidates with footprints extended to a ``kernelSize`` box.
Raises
------
RuntimeError
If ``candidateList`` is empty after sub-selection.
Definition at line 275 of file makeKernel.py.
276 candidateList, preconvolved=False):
277 """Make a list of acceptable KernelCandidates.
278
279 Generate a list of candidate sources for Psf-matching, remove sources
280 with bad pixel masks set or that extend off the image.
281
282 Parameters
283 ----------
284 convolved : `lsst.afw.image.Exposure`
285 Exposure that will be convolved. This is typically the template
286 image, and may have a large bbox than the reference exposure.
287 reference : `lsst.afw.image.Exposure`
288 Exposure that will be matched-to. This is typically the science
289 image.
290 kernelSize : `float`
291 Dimensions of the Psf-matching Kernel, used to set detection
292 footprints.
293 candidateList : `lsst.afw.table.SourceCatalog`
294 List of Sources to examine for kernel candidacy.
295 preconvolved : `bool`, optional
296 Was the science exposure already convolved with its PSF?
297
298 Returns
299 -------
300 candidates : `lsst.afw.table.SourceCatalog`
301 Candidates with footprints extended to a ``kernelSize`` box.
302
303 Raises
304 ------
305 RuntimeError
306 If ``candidateList`` is empty after sub-selection.
307 """
308 if candidateList is None:
309 candidateList = self.getSelectSources(reference, doSmooth=not preconvolved)
310 if len(candidateList) < 1:
311 raise RuntimeError("No kernel candidates after detection and measurement.")
312
313 bitmask = reference.mask.getPlaneBitMask(self.config.badMaskPlanes)
314 good = np.ones(len(candidateList), dtype=bool)
315 # Make all candidates have the same size footprint, based on kernelSize.
316 for i, candidate in enumerate(candidateList):
317 # Only use the brightest peak; the input should be pre-deblended!
318 peak = candidate.getFootprint().getPeaks()[0]
319 size = 2*kernelSize + 1 # ensure the resulting box is odd
320 bbox = lsst.geom.Box2I.makeCenteredBox(candidate.getCentroid(),
321 lsst.geom.Extent2I(size, size))
323 boxFootprint.addPeak(peak.getFx(), peak.getFy(), peak.getPeakValue())
324 candidate.setFootprint(boxFootprint)
325
326 # Reject footprints not contained in either image.
327 if not reference.getBBox().contains(bbox) or not convolved.getBBox().contains(bbox):
328 good[i] = False
329 continue
330 # Reject footprints with any bad mask bits set.
331 if (reference.subset(bbox).mask.array & bitmask).any():
332 good[i] = False
333 continue
334 candidates = candidateList[good].copy(deep=True)
335
336 self.log.info("Selected %d / %d sources as kernel candidates.", good.sum(), len(candidateList))
337
338 if len(candidates) < 1:
339 raise RuntimeError("No good kernel candidates available.")
340
341 return candidates
342
Class to describe the properties of a detected object from an image.
Definition Footprint.h:63
static Box2I makeCenteredBox(Point2D const ¢er, Extent const &size)
Create a box centered as closely as possible on a particular point.
Definition Box.cc:97
◆ makeKernelBasisList()
| lsst.ip.diffim.makeKernel.MakeKernelTask.makeKernelBasisList | ( | self, | |
| targetFwhmPix = None, | |||
| referenceFwhmPix = None, | |||
| basisDegGauss = None, | |||
| basisSigmaGauss = None, | |||
| metadata = None ) |
Wrapper to set log messages for
`lsst.ip.diffim.makeKernelBasisList`.
Parameters
----------
targetFwhmPix : `float`, optional
Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
Not used for delta function basis sets.
referenceFwhmPix : `float`, optional
Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
Not used for delta function basis sets.
basisDegGauss : `list` of `int`, optional
Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
Not used for delta function basis sets.
basisSigmaGauss : `list` of `int`, optional
Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
Not used for delta function basis sets.
metadata : `lsst.daf.base.PropertySet`, optional
Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
Not used for delta function basis sets.
Returns
-------
basisList: `list` of `lsst.afw.math.kernel.FixedKernel`
List of basis kernels.
Definition at line 343 of file makeKernel.py.
344 basisDegGauss=None, basisSigmaGauss=None, metadata=None):
345 """Wrapper to set log messages for
346 `lsst.ip.diffim.makeKernelBasisList`.
347
348 Parameters
349 ----------
350 targetFwhmPix : `float`, optional
351 Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
352 Not used for delta function basis sets.
353 referenceFwhmPix : `float`, optional
354 Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
355 Not used for delta function basis sets.
356 basisDegGauss : `list` of `int`, optional
357 Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
358 Not used for delta function basis sets.
359 basisSigmaGauss : `list` of `int`, optional
360 Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
361 Not used for delta function basis sets.
362 metadata : `lsst.daf.base.PropertySet`, optional
363 Passed on to `lsst.ip.diffim.generateAlardLuptonBasisList`.
364 Not used for delta function basis sets.
365
366 Returns
367 -------
368 basisList: `list` of `lsst.afw.math.kernel.FixedKernel`
369 List of basis kernels.
370 """
371 basisList = makeKernelBasisList(self.kConfig,
372 targetFwhmPix=targetFwhmPix,
373 referenceFwhmPix=referenceFwhmPix,
374 basisDegGauss=basisDegGauss,
375 basisSigmaGauss=basisSigmaGauss,
376 metadata=metadata)
377 if targetFwhmPix == referenceFwhmPix:
378 self.log.info("Target and reference psf fwhms are equal, falling back to config values")
379 elif referenceFwhmPix > targetFwhmPix:
380 self.log.info("Reference psf fwhm is the greater, normal convolution mode")
381 else:
382 self.log.info("Target psf fwhm is the greater, deconvolution mode")
383
384 return basisList
385
◆ run()
| lsst.ip.diffim.makeKernel.MakeKernelTask.run | ( | self, | |
| template, | |||
| science, | |||
| kernelSources, | |||
| preconvolved = False ) |
Solve for the kernel and background model that best match two
Exposures evaluated at the given source locations.
Parameters
----------
template : `lsst.afw.image.Exposure`
Exposure that will be convolved.
science : `lsst.afw.image.Exposure`
The exposure that will be matched.
kernelSources : `lsst.afw.table.SourceCatalog`
Kernel candidate sources with appropriately sized footprints.
Typically the output of `MakeKernelTask.selectKernelSources`.
preconvolved : `bool`, optional
Was the science image convolved with its own PSF?
Returns
-------
results : `lsst.pipe.base.Struct`
``psfMatchingKernel`` : `lsst.afw.math.LinearCombinationKernel`
Spatially varying Psf-matching kernel.
``backgroundModel`` : `lsst.afw.math.Function2D`
Spatially varying background-matching function.
Definition at line 108 of file makeKernel.py.
108 def run(self, template, science, kernelSources, preconvolved=False):
109 """Solve for the kernel and background model that best match two
110 Exposures evaluated at the given source locations.
111
112 Parameters
113 ----------
114 template : `lsst.afw.image.Exposure`
115 Exposure that will be convolved.
116 science : `lsst.afw.image.Exposure`
117 The exposure that will be matched.
118 kernelSources : `lsst.afw.table.SourceCatalog`
119 Kernel candidate sources with appropriately sized footprints.
120 Typically the output of `MakeKernelTask.selectKernelSources`.
121 preconvolved : `bool`, optional
122 Was the science image convolved with its own PSF?
123
124 Returns
125 -------
126 results : `lsst.pipe.base.Struct`
127
128 ``psfMatchingKernel`` : `lsst.afw.math.LinearCombinationKernel`
129 Spatially varying Psf-matching kernel.
130 ``backgroundModel`` : `lsst.afw.math.Function2D`
131 Spatially varying background-matching function.
132 """
133 kernelCellSet = self._buildCellSet(template.maskedImage, science.maskedImage, kernelSources)
134 # Calling getPsfFwhm on template.psf fails on some rare occasions when
135 # the template has no input exposures at the average position of the
136 # stars. So we try getPsfFwhm first on template, and if that fails we
137 # evaluate the PSF on a grid specified by fwhmExposure* fields.
138 # To keep consistent definitions for PSF size on the template and
139 # science images, we use the same method for both.
140 try:
141 templateFwhmPix = getPsfFwhm(template.psf)
142 scienceFwhmPix = getPsfFwhm(science.psf)
143 except InvalidParameterError:
144 self.log.debug("Unable to evaluate PSF at the average position. "
145 "Evaluting PSF on a grid of points."
146 )
147 templateFwhmPix = evaluateMeanPsfFwhm(template,
148 fwhmExposureBuffer=self.config.fwhmExposureBuffer,
149 fwhmExposureGrid=self.config.fwhmExposureGrid
150 )
151 scienceFwhmPix = evaluateMeanPsfFwhm(science,
152 fwhmExposureBuffer=self.config.fwhmExposureBuffer,
153 fwhmExposureGrid=self.config.fwhmExposureGrid
154 )
155
156 if preconvolved:
157 scienceFwhmPix *= np.sqrt(2)
158 basisList = self.makeKernelBasisList(templateFwhmPix, scienceFwhmPix,
159 metadata=self.metadata)
160 spatialSolution, psfMatchingKernel, backgroundModel = self._solve(kernelCellSet, basisList)
161 return lsst.pipe.base.Struct(
162 psfMatchingKernel=psfMatchingKernel,
163 backgroundModel=backgroundModel,
164 )
165
◆ selectKernelSources()
| lsst.ip.diffim.makeKernel.MakeKernelTask.selectKernelSources | ( | self, | |
| template, | |||
| science, | |||
| candidateList = None, | |||
| preconvolved = False ) |
Select sources from a list of candidates, and extract footprints.
Parameters
----------
template : `lsst.afw.image.Exposure`
Exposure that will be convolved.
science : `lsst.afw.image.Exposure`
The exposure that will be matched.
candidateList : `lsst.afw.table.SourceCatalog`
Sources to check as possible kernel candidates.
preconvolved : `bool`, optional
Was the science image convolved with its own PSF?
Returns
-------
kernelSources : `lsst.afw.table.SourceCatalog`
Kernel candidates with appropriate sized footprints.
Definition at line 166 of file makeKernel.py.
166 def selectKernelSources(self, template, science, candidateList=None, preconvolved=False):
167 """Select sources from a list of candidates, and extract footprints.
168
169 Parameters
170 ----------
171 template : `lsst.afw.image.Exposure`
172 Exposure that will be convolved.
173 science : `lsst.afw.image.Exposure`
174 The exposure that will be matched.
175 candidateList : `lsst.afw.table.SourceCatalog`
176 Sources to check as possible kernel candidates.
177 preconvolved : `bool`, optional
178 Was the science image convolved with its own PSF?
179
180 Returns
181 -------
182 kernelSources : `lsst.afw.table.SourceCatalog`
183 Kernel candidates with appropriate sized footprints.
184 """
185 # Calling getPsfFwhm on template.psf fails on some rare occasions when
186 # the template has no input exposures at the average position of the
187 # stars. So we try getPsfFwhm first on template, and if that fails we
188 # evaluate the PSF on a grid specified by fwhmExposure* fields.
189 # To keep consistent definitions for PSF size on the template and
190 # science images, we use the same method for both.
191 try:
192 templateFwhmPix = getPsfFwhm(template.psf)
193 scienceFwhmPix = getPsfFwhm(science.psf)
194 except InvalidParameterError:
195 self.log.debug("Unable to evaluate PSF at the average position. "
196 "Evaluting PSF on a grid of points."
197 )
198 templateFwhmPix = evaluateMeanPsfFwhm(template,
199 fwhmExposureBuffer=self.config.fwhmExposureBuffer,
200 fwhmExposureGrid=self.config.fwhmExposureGrid
201 )
202 scienceFwhmPix = evaluateMeanPsfFwhm(science,
203 fwhmExposureBuffer=self.config.fwhmExposureBuffer,
204 fwhmExposureGrid=self.config.fwhmExposureGrid
205 )
206 if preconvolved:
207 scienceFwhmPix *= np.sqrt(2)
208 kernelSize = self.makeKernelBasisList(templateFwhmPix, scienceFwhmPix)[0].getWidth()
209 kernelSources = self.makeCandidateList(template, science, kernelSize,
210 candidateList=candidateList,
211 preconvolved=preconvolved)
212 return kernelSources
213
Member Data Documentation
◆ _DefaultName
|
staticprotected |
Definition at line 94 of file makeKernel.py.
◆ background
| lsst.ip.diffim.makeKernel.MakeKernelTask.background |
Definition at line 101 of file makeKernel.py.
◆ ConfigClass
|
static |
Definition at line 93 of file makeKernel.py.
◆ kConfig
| lsst.ip.diffim.makeKernel.MakeKernelTask.kConfig |
Definition at line 98 of file makeKernel.py.
◆ selectAlgMetadata
| lsst.ip.diffim.makeKernel.MakeKernelTask.selectAlgMetadata |
Definition at line 104 of file makeKernel.py.
◆ selectSchema
| lsst.ip.diffim.makeKernel.MakeKernelTask.selectSchema |
Definition at line 103 of file makeKernel.py.
The documentation for this class was generated from the following file:
- /j/snowflake/release/lsstsw/stack/lsst-scipipe-8.0.0/Linux64/ip_diffim/g1470d8bcf6+190ad2ba91/python/lsst/ip/diffim/makeKernel.py
