Coverage for python/lsst/meas/extensions/piff/piffPsfDeterminer.py: 21%
112 statements
« prev ^ index » next coverage.py v6.5.0, created at 2022-11-09 07:00 +0000
« prev ^ index » next coverage.py v6.5.0, created at 2022-11-09 07:00 +0000
1# This file is part of meas_extensions_piff.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (https://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This program is free software: you can redistribute it and/or modify
10# it under the terms of the GNU General Public License as published by
11# the Free Software Foundation, either version 3 of the License, or
12# (at your option) any later version.
13#
14# This program is distributed in the hope that it will be useful,
15# but WITHOUT ANY WARRANTY; without even the implied warranty of
16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17# GNU General Public License for more details.
18#
19# You should have received a copy of the GNU General Public License
20# along with this program. If not, see <https://www.gnu.org/licenses/>.
22__all__ = ["PiffPsfDeterminerConfig", "PiffPsfDeterminerTask"]
24import numpy as np
25import piff
26import galsim
27import re
29import lsst.pex.config as pexConfig
30import lsst.meas.algorithms as measAlg
31from lsst.meas.algorithms.psfDeterminer import BasePsfDeterminerTask
32from .piffPsf import PiffPsf
35def _validateGalsimInterpolant(name: str) -> bool:
36 """A helper function to validate the GalSim interpolant at config time.
38 Parameters
39 ----------
40 name : str
41 The name of the interpolant to use from GalSim. Valid options are:
42 galsim.Lanczos(N) or Lancsos(N), where N is a positive integer
43 galsim.Linear
44 galsim.Cubic
45 galsim.Quintic
46 galsim.Delta
47 galsim.Nearest
48 galsim.SincInterpolant
50 Returns
51 -------
52 is_valid : bool
53 Whether the provided interpolant name is valid.
54 """
55 # First, check if ``name`` is a valid Lanczos interpolant.
56 for pattern in (re.compile(r"Lanczos\(\d+\)"), re.compile(r"galsim.Lanczos\(\d+\)"),):
57 match = re.match(pattern, name) # Search from the start of the string.
58 if match is not None:
59 # Check that the pattern is also the end of the string.
60 return match.end() == len(name)
62 # If not, check if ``name`` is any other valid GalSim interpolant.
63 names = {f"galsim.{interp}" for interp in
64 ("Cubic", "Delta", "Linear", "Nearest", "Quintic", "SincInterpolant")
65 }
66 return name in names
69class PiffPsfDeterminerConfig(BasePsfDeterminerTask.ConfigClass):
70 spatialOrder = pexConfig.Field(
71 doc="specify spatial order for PSF kernel creation",
72 dtype=int,
73 default=2,
74 )
75 samplingSize = pexConfig.Field(
76 doc="Resolution of the internal PSF model relative to the pixel size; "
77 "e.g. 0.5 is equal to 2x oversampling",
78 dtype=float,
79 default=1,
80 )
81 outlierNSigma = pexConfig.Field(
82 doc="n sigma for chisq outlier rejection",
83 dtype=float,
84 default=4.0
85 )
86 outlierMaxRemove = pexConfig.Field(
87 doc="Max fraction of stars to remove as outliers each iteration",
88 dtype=float,
89 default=0.05
90 )
91 maxSNR = pexConfig.Field(
92 doc="Rescale the weight of bright stars such that their SNR is less "
93 "than this value.",
94 dtype=float,
95 default=200.0
96 )
97 zeroWeightMaskBits = pexConfig.ListField(
98 doc="List of mask bits for which to set pixel weights to zero.",
99 dtype=str,
100 default=['BAD', 'CR', 'INTRP', 'SAT', 'SUSPECT', 'NO_DATA']
101 )
102 minimumUnmaskedFraction = pexConfig.Field(
103 doc="Minimum fraction of unmasked pixels required to use star.",
104 dtype=float,
105 default=0.5
106 )
107 interpolant = pexConfig.Field(
108 doc="GalSim interpolant name for Piff to use. "
109 "Options include 'Lanczos(N)', where N is an integer, along with "
110 "galsim.Cubic, galsim.Delta, galsim.Linear, galsim.Nearest, "
111 "galsim.Quintic, and galsim.SincInterpolant.",
112 dtype=str,
113 check=_validateGalsimInterpolant,
114 default="Lanczos(11)",
115 )
117 def setDefaults(self):
118 # kernelSize should be at least 25 so that
119 # i) aperture flux with 12 pixel radius can be compared to PSF flux.
120 # ii) fake sources injected to match the 12 pixel aperture flux get
121 # measured correctly
122 self.kernelSize = 25
123 self.kernelSizeMin = 11
124 self.kernelSizeMax = 35
127def getGoodPixels(maskedImage, zeroWeightMaskBits):
128 """Compute an index array indicating good pixels to use.
130 Parameters
131 ----------
132 maskedImage : `afw.image.MaskedImage`
133 PSF candidate postage stamp
134 zeroWeightMaskBits : `List[str]`
135 List of mask bits for which to set pixel weights to zero.
137 Returns
138 -------
139 good : `ndarray`
140 Index array indicating good pixels.
141 """
142 imArr = maskedImage.image.array
143 varArr = maskedImage.variance.array
144 bitmask = maskedImage.mask.getPlaneBitMask(zeroWeightMaskBits)
145 good = (
146 (varArr != 0)
147 & (np.isfinite(varArr))
148 & (np.isfinite(imArr))
149 & ((maskedImage.mask.array & bitmask) == 0)
150 )
151 return good
154def computeWeight(maskedImage, maxSNR, good):
155 """Derive a weight map without Poisson variance component due to signal.
157 Parameters
158 ----------
159 maskedImage : `afw.image.MaskedImage`
160 PSF candidate postage stamp
161 maxSNR : `float`
162 Maximum SNR applying variance floor.
163 good : `ndarray`
164 Index array indicating good pixels.
166 Returns
167 -------
168 weightArr : `ndarry`
169 Array to use for weight.
170 """
171 imArr = maskedImage.image.array
172 varArr = maskedImage.variance.array
174 # Fit a straight line to variance vs (sky-subtracted) signal.
175 # The evaluate that line at zero signal to get an estimate of the
176 # signal-free variance.
177 fit = np.polyfit(imArr[good], varArr[good], deg=1)
178 # fit is [1/gain, sky_var]
179 weightArr = np.zeros_like(imArr, dtype=float)
180 weightArr[good] = 1./fit[1]
182 applyMaxSNR(imArr, weightArr, good, maxSNR)
183 return weightArr
186def applyMaxSNR(imArr, weightArr, good, maxSNR):
187 """Rescale weight of bright stars to cap the computed SNR.
189 Parameters
190 ----------
191 imArr : `ndarray`
192 Signal (image) array of stamp.
193 weightArr : `ndarray`
194 Weight map array. May be rescaled in place.
195 good : `ndarray`
196 Index array of pixels to use when computing SNR.
197 maxSNR : `float`
198 Threshold for adjusting variance plane implementing maximum SNR.
199 """
200 # We define the SNR value following Piff. Here's the comment from that
201 # code base explaining the calculation.
202 #
203 # The S/N value that we use will be the weighted total flux where the
204 # weight function is the star's profile itself. This is the maximum S/N
205 # value that any flux measurement can possibly produce, which will be
206 # closer to an in-practice S/N than using all the pixels equally.
207 #
208 # F = Sum_i w_i I_i^2
209 # var(F) = Sum_i w_i^2 I_i^2 var(I_i)
210 # = Sum_i w_i I_i^2 <--- Assumes var(I_i) = 1/w_i
211 #
212 # S/N = F / sqrt(var(F))
213 #
214 # Note that if the image is pure noise, this will produce a "signal" of
215 #
216 # F_noise = Sum_i w_i 1/w_i = Npix
217 #
218 # So for a more accurate estimate of the S/N of the actual star itself, one
219 # should subtract off Npix from the measured F.
220 #
221 # The final formula then is:
222 #
223 # F = Sum_i w_i I_i^2
224 # S/N = (F-Npix) / sqrt(F)
225 F = np.sum(weightArr[good]*imArr[good]**2, dtype=float)
226 Npix = np.sum(good)
227 SNR = 0.0 if F < Npix else (F-Npix)/np.sqrt(F)
228 # rescale weight of bright stars. Essentially makes an error floor.
229 if SNR > maxSNR:
230 factor = (maxSNR / SNR)**2
231 weightArr[good] *= factor
234def _computeWeightAlternative(maskedImage, maxSNR):
235 """Alternative algorithm for creating weight map.
237 This version is equivalent to that used by Piff internally. The weight map
238 it produces tends to leave a residual when removing the Poisson component
239 due to the signal. We leave it here as a reference, but without intending
240 that it be used (or be maintained).
241 """
242 imArr = maskedImage.image.array
243 varArr = maskedImage.variance.array
244 good = (varArr != 0) & np.isfinite(varArr) & np.isfinite(imArr)
246 fit = np.polyfit(imArr[good], varArr[good], deg=1)
247 # fit is [1/gain, sky_var]
248 gain = 1./fit[0]
249 varArr[good] -= imArr[good] / gain
250 weightArr = np.zeros_like(imArr, dtype=float)
251 weightArr[good] = 1./varArr[good]
253 applyMaxSNR(imArr, weightArr, good, maxSNR)
254 return weightArr
257class PiffPsfDeterminerTask(BasePsfDeterminerTask):
258 """A measurePsfTask PSF estimator using Piff as the implementation.
259 """
260 ConfigClass = PiffPsfDeterminerConfig
261 _DefaultName = "psfDeterminer.Piff"
263 def determinePsf(
264 self, exposure, psfCandidateList, metadata=None, flagKey=None
265 ):
266 """Determine a Piff PSF model for an exposure given a list of PSF
267 candidates.
269 Parameters
270 ----------
271 exposure : `lsst.afw.image.Exposure`
272 Exposure containing the PSF candidates.
273 psfCandidateList : `list` of `lsst.meas.algorithms.PsfCandidate`
274 A sequence of PSF candidates typically obtained by detecting sources
275 and then running them through a star selector.
276 metadata : `lsst.daf.base import PropertyList` or `None`, optional
277 A home for interesting tidbits of information.
278 flagKey : `str` or `None`, optional
279 Schema key used to mark sources actually used in PSF determination.
281 Returns
282 -------
283 psf : `lsst.meas.extensions.piff.PiffPsf`
284 The measured PSF model.
285 psfCellSet : `None`
286 Unused by this PsfDeterminer.
287 """
288 kernelSize = int(np.clip(
289 self.config.kernelSize,
290 self.config.kernelSizeMin,
291 self.config.kernelSizeMax
292 ))
293 self._validatePsfCandidates(psfCandidateList, kernelSize, self.config.samplingSize)
295 stars = []
296 for candidate in psfCandidateList:
297 cmi = candidate.getMaskedImage()
298 good = getGoodPixels(cmi, self.config.zeroWeightMaskBits)
299 fracGood = np.sum(good)/good.size
300 if fracGood < self.config.minimumUnmaskedFraction:
301 continue
302 weight = computeWeight(cmi, self.config.maxSNR, good)
304 bbox = cmi.getBBox()
305 bds = galsim.BoundsI(
306 galsim.PositionI(*bbox.getMin()),
307 galsim.PositionI(*bbox.getMax())
308 )
309 gsImage = galsim.Image(bds, scale=1.0, dtype=float)
310 gsImage.array[:] = cmi.image.array
311 gsWeight = galsim.Image(bds, scale=1.0, dtype=float)
312 gsWeight.array[:] = weight
314 source = candidate.getSource()
315 image_pos = galsim.PositionD(source.getX(), source.getY())
317 data = piff.StarData(
318 gsImage,
319 image_pos,
320 weight=gsWeight
321 )
322 stars.append(piff.Star(data, None))
324 piffConfig = {
325 'type': "Simple",
326 'model': {
327 'type': 'PixelGrid',
328 'scale': self.config.samplingSize,
329 'size': kernelSize,
330 'interp': self.config.interpolant
331 },
332 'interp': {
333 'type': 'BasisPolynomial',
334 'order': self.config.spatialOrder
335 },
336 'outliers': {
337 'type': 'Chisq',
338 'nsigma': self.config.outlierNSigma,
339 'max_remove': self.config.outlierMaxRemove
340 }
341 }
343 piffResult = piff.PSF.process(piffConfig)
344 # Run on a single CCD, and in image coords rather than sky coords.
345 wcs = {0: galsim.PixelScale(1.0)}
346 pointing = None
348 piffResult.fit(stars, wcs, pointing, logger=self.log)
349 drawSize = 2*np.floor(0.5*kernelSize/self.config.samplingSize) + 1
350 psf = PiffPsf(drawSize, drawSize, piffResult)
352 used_image_pos = [s.image_pos for s in piffResult.stars]
353 if flagKey:
354 for candidate in psfCandidateList:
355 source = candidate.getSource()
356 posd = galsim.PositionD(source.getX(), source.getY())
357 if posd in used_image_pos:
358 source.set(flagKey, True)
360 if metadata is not None:
361 metadata["spatialFitChi2"] = piffResult.chisq
362 metadata["numAvailStars"] = len(stars)
363 metadata["numGoodStars"] = len(piffResult.stars)
364 metadata["avgX"] = np.mean([p.x for p in piffResult.stars])
365 metadata["avgY"] = np.mean([p.y for p in piffResult.stars])
367 return psf, None
369 def _validatePsfCandidates(self, psfCandidateList, kernelSize, samplingSize):
370 """Raise if psfCandidates are smaller than the configured kernelSize.
372 Parameters
373 ----------
374 psfCandidateList : `list` of `lsst.meas.algorithms.PsfCandidate`
375 Sequence of psf candidates to check.
376 kernelSize : `int`
377 Size of image model to use in PIFF.
378 samplingSize : `float`
379 Resolution of the internal PSF model relative to the pixel size.
381 Raises
382 ------
383 RuntimeError
384 Raised if any psfCandidate has width or height smaller than
385 config.kernelSize.
386 """
387 # We can assume all candidates have the same dimensions.
388 candidate = psfCandidateList[0]
389 drawSize = int(2*np.floor(0.5*kernelSize/samplingSize) + 1)
390 if (candidate.getHeight() < drawSize
391 or candidate.getWidth() < drawSize):
392 raise RuntimeError("PSF candidates must be at least config.kernelSize/config.samplingSize="
393 f"{drawSize} pixels per side; "
394 f"found {candidate.getWidth()}x{candidate.getHeight()}.")
397measAlg.psfDeterminerRegistry.register("piff", PiffPsfDeterminerTask)