Coverage for python/lsst/meas/algorithms/dynamicDetection.py: 16%
178 statements
« prev ^ index » next coverage.py v6.5.0, created at 2024-03-20 01:03 -0700
« prev ^ index » next coverage.py v6.5.0, created at 2024-03-20 01:03 -0700
2__all__ = ["DynamicDetectionConfig", "DynamicDetectionTask"]
4import numpy as np
6from lsst.pex.config import Field, ConfigurableField
7from lsst.pipe.base import Struct, NoWorkFound
9from .detection import SourceDetectionConfig, SourceDetectionTask
10from .skyObjects import SkyObjectsTask
12from lsst.afw.detection import FootprintSet
13from lsst.afw.geom import makeCdMatrix, makeSkyWcs
14from lsst.afw.table import SourceCatalog, SourceTable
15from lsst.meas.base import ForcedMeasurementTask
17import lsst.afw.image
18import lsst.afw.math
19import lsst.geom as geom
22class DynamicDetectionConfig(SourceDetectionConfig):
23 """Configuration for DynamicDetectionTask
24 """
25 prelimThresholdFactor = Field(dtype=float, default=0.5,
26 doc="Factor by which to multiply the main detection threshold "
27 "(thresholdValue) to use for first pass (to find sky objects).")
28 prelimNegMultiplier = Field(dtype=float, default=2.5,
29 doc="Multiplier for the negative (relative to positive) polarity "
30 "detections threshold to use for first pass (to find sky objects).")
31 skyObjects = ConfigurableField(target=SkyObjectsTask, doc="Generate sky objects.")
32 doBackgroundTweak = Field(dtype=bool, default=True,
33 doc="Tweak background level so median PSF flux of sky objects is zero?")
34 minFractionSources = Field(dtype=float, default=0.02,
35 doc="Minimum fraction of the requested number of sky sources for dynamic "
36 "detection to be considered a success. If the number of good sky sources "
37 "identified falls below this threshold, a NoWorkFound error is raised so "
38 "that this dataId is no longer considered in downstream processing.")
39 doBrightPrelimDetection = Field(dtype=bool, default=True,
40 doc="Do initial bright detection pass where footprints are grown "
41 "by brightGrowFactor?")
42 brightMultiplier = Field(dtype=float, default=2000.0,
43 doc="Multiplier to apply to the prelimThresholdFactor for the "
44 "\"bright\" detections stage (want this to be large to only "
45 "detect the brightest sources).")
46 brightNegFactor = Field(dtype=float, default=2.2,
47 doc="Factor by which to multiply the threshold for the negative polatiry "
48 "detections for the \"bright\" detections stage (this needs to be fairly "
49 "low given the nature of the negative polarity detections in the very "
50 "large positive polarity threshold).")
51 brightGrowFactor = Field(dtype=int, default=40,
52 doc="Factor by which to grow the footprints of sources detected in the "
53 "\"bright\" detections stage (want this to be large to mask wings of "
54 "bright sources).")
55 brightMaskFractionMax = Field(dtype=float, default=0.95,
56 doc="Maximum allowed fraction of masked pixes from the \"bright\" "
57 "detection stage (to mask regions unsuitable for sky sourcess). "
58 "If this fraction is exeeded, the detection threshold for this stage "
59 "will be increased by bisectFactor until the fraction of masked "
60 "pixels drops below this threshold.")
61 bisectFactor = Field(dtype=float, default=1.2,
62 doc="Factor by which to increase thresholds in brightMaskFractionMax loop.")
64 def setDefaults(self):
65 SourceDetectionConfig.setDefaults(self)
66 self.skyObjects.nSources = 1000 # For good statistics
67 for maskStr in ["SAT"]:
68 if maskStr not in self.skyObjects.avoidMask:
69 self.skyObjects.avoidMask.append(maskStr)
72class DynamicDetectionTask(SourceDetectionTask):
73 """Detection of sources on an image with a dynamic threshold
75 We first detect sources using a lower threshold than normal (see config
76 parameter ``prelimThresholdFactor``) in order to identify good sky regions
77 (configurable ``skyObjects``). Then we perform forced PSF photometry on
78 those sky regions. Using those PSF flux measurements and estimated errors,
79 we set the threshold so that the stdev of the measurements matches the
80 median estimated error.
82 Besides the usual initialisation of configurables, we also set up
83 the forced measurement which is deliberately not represented in
84 this Task's configuration parameters because we're using it as
85 part of the algorithm and we don't want to allow it to be modified.
86 """
87 ConfigClass = DynamicDetectionConfig
88 _DefaultName = "dynamicDetection"
90 def __init__(self, *args, **kwargs):
92 SourceDetectionTask.__init__(self, *args, **kwargs)
93 self.makeSubtask("skyObjects")
95 # Set up forced measurement.
96 config = ForcedMeasurementTask.ConfigClass()
97 config.plugins.names = ['base_TransformedCentroid', 'base_PsfFlux', 'base_LocalBackground']
98 # We'll need the "centroid" and "psfFlux" slots
99 for slot in ("shape", "psfShape", "apFlux", "modelFlux", "gaussianFlux", "calibFlux"):
100 setattr(config.slots, slot, None)
101 config.copyColumns = {}
102 self.skySchema = SourceTable.makeMinimalSchema()
103 self.skyMeasurement = ForcedMeasurementTask(config=config, name="skyMeasurement", parentTask=self,
104 refSchema=self.skySchema)
106 def calculateThreshold(self, exposure, seed, sigma=None, minFractionSourcesFactor=1.0, isBgTweak=False):
107 """Calculate new threshold
109 This is the main functional addition to the vanilla
110 `SourceDetectionTask`.
112 We identify sky objects and perform forced PSF photometry on
113 them. Using those PSF flux measurements and estimated errors,
114 we set the threshold so that the stdev of the measurements
115 matches the median estimated error.
117 Parameters
118 ----------
119 exposure : `lsst.afw.image.Exposure`
120 Exposure on which we're detecting sources.
121 seed : `int`
122 RNG seed to use for finding sky objects.
123 sigma : `float`, optional
124 Gaussian sigma of smoothing kernel; if not provided,
125 will be deduced from the exposure's PSF.
126 minFractionSourcesFactor : `float`
127 Change the fraction of required sky sources from that set in
128 ``self.config.minFractionSources`` by this factor. NOTE: this
129 is intended for use in the background tweak pass (the detection
130 threshold is much lower there, so many more pixels end up marked
131 as DETECTED or DETECTED_NEGATIVE, leaving less room for sky
132 object placement).
133 isBgTweak : `bool`
134 Set to ``True`` for the background tweak pass (for more helpful
135 log messages).
137 Returns
138 -------
139 result : `lsst.pipe.base.Struct`
140 Result struct with components:
142 ``multiplicative``
143 Multiplicative factor to be applied to the
144 configured detection threshold (`float`).
145 ``additive``
146 Additive factor to be applied to the background
147 level (`float`).
149 Raises
150 ------
151 NoWorkFound
152 Raised if the number of good sky sources found is less than the
153 minimum fraction
154 (``self.config.minFractionSources``*``minFractionSourcesFactor``)
155 of the number requested (``self.skyObjects.config.nSources``).
156 """
157 wcsIsNone = exposure.getWcs() is None
158 if wcsIsNone: # create a dummy WCS as needed by ForcedMeasurementTask
159 self.log.info("WCS for exposure is None. Setting a dummy WCS for dynamic detection.")
160 exposure.setWcs(makeSkyWcs(crpix=geom.Point2D(0, 0),
161 crval=geom.SpherePoint(0, 0, geom.degrees),
162 cdMatrix=makeCdMatrix(scale=1e-5*geom.degrees)))
163 fp = self.skyObjects.run(exposure.maskedImage.mask, seed)
164 skyFootprints = FootprintSet(exposure.getBBox())
165 skyFootprints.setFootprints(fp)
166 table = SourceTable.make(self.skyMeasurement.schema)
167 catalog = SourceCatalog(table)
168 catalog.reserve(len(skyFootprints.getFootprints()))
169 skyFootprints.makeSources(catalog)
170 key = catalog.getCentroidSlot().getMeasKey()
171 for source in catalog:
172 peaks = source.getFootprint().getPeaks()
173 assert len(peaks) == 1
174 source.set(key, peaks[0].getF())
175 source.updateCoord(exposure.getWcs())
177 # Forced photometry on sky objects
178 self.skyMeasurement.run(catalog, exposure, catalog, exposure.getWcs())
180 # Calculate new threshold
181 fluxes = catalog["base_PsfFlux_instFlux"]
182 area = catalog["base_PsfFlux_area"]
183 bg = catalog["base_LocalBackground_instFlux"]
185 good = (~catalog["base_PsfFlux_flag"] & ~catalog["base_LocalBackground_flag"]
186 & np.isfinite(fluxes) & np.isfinite(area) & np.isfinite(bg))
188 minNumSources = int(self.config.minFractionSources*self.skyObjects.config.nSources)
189 # Reduce the number of sky sources required if requested, but ensure
190 # a minumum of 3.
191 if minFractionSourcesFactor != 1.0:
192 minNumSources = max(3, int(minNumSources*minFractionSourcesFactor))
193 if good.sum() < minNumSources:
194 if not isBgTweak:
195 msg = (f"Insufficient good sky source flux measurements ({good.sum()} < "
196 f"{minNumSources}) for dynamic threshold calculation.")
197 else:
198 msg = (f"Insufficient good sky source flux measurements ({good.sum()} < "
199 f"{minNumSources}) for background tweak calculation.")
201 nPix = exposure.mask.array.size
202 badPixelMask = lsst.afw.image.Mask.getPlaneBitMask(["NO_DATA", "BAD"])
203 nGoodPix = np.sum(exposure.mask.array & badPixelMask == 0)
204 if nGoodPix/nPix > 0.2:
205 detectedPixelMask = lsst.afw.image.Mask.getPlaneBitMask(["DETECTED", "DETECTED_NEGATIVE"])
206 nDetectedPix = np.sum(exposure.mask.array & detectedPixelMask != 0)
207 msg += (f" However, {nGoodPix}/{nPix} pixels are not marked NO_DATA or BAD, "
208 "so there should be sufficient area to locate suitable sky sources. "
209 f"Note that {nDetectedPix} of {nGoodPix} \"good\" pixels were marked "
210 "as DETECTED or DETECTED_NEGATIVE.")
211 raise RuntimeError(msg)
212 raise NoWorkFound(msg)
214 if not isBgTweak:
215 self.log.info("Number of good sky sources used for dynamic detection: %d (of %d requested).",
216 good.sum(), self.skyObjects.config.nSources)
217 else:
218 self.log.info("Number of good sky sources used for dynamic detection background tweak:"
219 " %d (of %d requested).", good.sum(), self.skyObjects.config.nSources)
220 bgMedian = np.median((fluxes/area)[good])
222 lq, uq = np.percentile((fluxes - bg*area)[good], [25.0, 75.0])
223 stdevMeas = 0.741*(uq - lq)
224 medianError = np.median(catalog["base_PsfFlux_instFluxErr"][good])
225 if wcsIsNone:
226 exposure.setWcs(None)
227 return Struct(multiplicative=medianError/stdevMeas, additive=bgMedian)
229 def detectFootprints(self, exposure, doSmooth=True, sigma=None, clearMask=True, expId=None):
230 """Detect footprints with a dynamic threshold
232 This varies from the vanilla ``detectFootprints`` method because we
233 do detection three times: first with a high threshold to detect
234 "bright" (both positive and negative, the latter to identify very
235 over-subtracted regions) sources for which we grow the DETECTED and
236 DETECTED_NEGATIVE masks significantly to account for wings. Second,
237 with a low threshold to mask all non-empty regions of the image. These
238 two masks are combined and used to identify regions of sky
239 uncontaminated by objects. A final round of detection is then done
240 with the new calculated threshold.
242 Parameters
243 ----------
244 exposure : `lsst.afw.image.Exposure`
245 Exposure to process; DETECTED{,_NEGATIVE} mask plane will be
246 set in-place.
247 doSmooth : `bool`, optional
248 If True, smooth the image before detection using a Gaussian
249 of width ``sigma``.
250 sigma : `float`, optional
251 Gaussian Sigma of PSF (pixels); used for smoothing and to grow
252 detections; if `None` then measure the sigma of the PSF of the
253 ``exposure``.
254 clearMask : `bool`, optional
255 Clear both DETECTED and DETECTED_NEGATIVE planes before running
256 detection.
257 expId : `int`, optional
258 Exposure identifier, used as a seed for the random number
259 generator. If absent, the seed will be the sum of the image.
261 Returns
262 -------
263 resutls : `lsst.pipe.base.Struct`
264 The results `~lsst.pipe.base.Struct` contains:
266 ``positive``
267 Positive polarity footprints.
268 (`lsst.afw.detection.FootprintSet` or `None`)
269 ``negative``
270 Negative polarity footprints.
271 (`lsst.afw.detection.FootprintSet` or `None`)
272 ``numPos``
273 Number of footprints in positive or 0 if detection polarity was
274 negative. (`int`)
275 ``numNeg``
276 Number of footprints in negative or 0 if detection polarity was
277 positive. (`int`)
278 ``background``
279 Re-estimated background. `None` if
280 ``reEstimateBackground==False``.
281 (`lsst.afw.math.BackgroundList`)
282 ``factor``
283 Multiplication factor applied to the configured detection
284 threshold. (`float`)
285 ``prelim``
286 Results from preliminary detection pass.
287 (`lsst.pipe.base.Struct`)
288 """
289 maskedImage = exposure.maskedImage
291 if clearMask:
292 self.clearMask(maskedImage.mask)
293 else:
294 oldDetected = maskedImage.mask.array & maskedImage.mask.getPlaneBitMask(["DETECTED",
295 "DETECTED_NEGATIVE"])
296 nPix = maskedImage.mask.array.size
297 badPixelMask = lsst.afw.image.Mask.getPlaneBitMask(["NO_DATA", "BAD"])
298 nGoodPix = np.sum(maskedImage.mask.array & badPixelMask == 0)
299 self.log.info("Number of good data pixels (i.e. not NO_DATA or BAD): {} ({:.1f}% of total)".
300 format(nGoodPix, 100*nGoodPix/nPix))
302 with self.tempWideBackgroundContext(exposure):
303 # Could potentially smooth with a wider kernel than the PSF in
304 # order to better pick up the wings of stars and galaxies, but for
305 # now sticking with the PSF as that's more simple.
306 psf = self.getPsf(exposure, sigma=sigma)
307 convolveResults = self.convolveImage(maskedImage, psf, doSmooth=doSmooth)
309 if self.config.doBrightPrelimDetection:
310 brightDetectedMask = self._computeBrightDetectionMask(maskedImage, convolveResults)
312 middle = convolveResults.middle
313 sigma = convolveResults.sigma
314 prelim = self.applyThreshold(
315 middle, maskedImage.getBBox(), factor=self.config.prelimThresholdFactor,
316 factorNeg=self.config.prelimNegMultiplier*self.config.prelimThresholdFactor
317 )
318 self.finalizeFootprints(
319 maskedImage.mask, prelim, sigma, factor=self.config.prelimThresholdFactor,
320 factorNeg=self.config.prelimNegMultiplier*self.config.prelimThresholdFactor
321 )
322 if self.config.doBrightPrelimDetection:
323 # Combine prelim and bright detection masks for multiplier
324 # determination.
325 maskedImage.mask.array |= brightDetectedMask
327 # Calculate the proper threshold
328 # seed needs to fit in a C++ 'int' so pybind doesn't choke on it
329 seed = (expId if expId is not None else int(maskedImage.image.array.sum())) % (2**31 - 1)
330 threshResults = self.calculateThreshold(exposure, seed, sigma=sigma)
331 factor = threshResults.multiplicative
332 self.log.info("Modifying configured detection threshold by factor %f to %f",
333 factor, factor*self.config.thresholdValue)
335 # Blow away preliminary (low threshold) detection mask
336 self.clearMask(maskedImage.mask)
337 if not clearMask:
338 maskedImage.mask.array |= oldDetected
340 # Rinse and repeat thresholding with new calculated threshold
341 results = self.applyThreshold(middle, maskedImage.getBBox(), factor)
342 results.prelim = prelim
343 results.background = lsst.afw.math.BackgroundList()
344 if self.config.doTempLocalBackground:
345 self.applyTempLocalBackground(exposure, middle, results)
346 self.finalizeFootprints(maskedImage.mask, results, sigma, factor=factor)
348 self.clearUnwantedResults(maskedImage.mask, results)
350 if self.config.reEstimateBackground:
351 self.reEstimateBackground(maskedImage, results.background)
353 self.display(exposure, results, middle)
355 if self.config.doBackgroundTweak:
356 # Re-do the background tweak after any temporary backgrounds have
357 # been restored.
358 #
359 # But we want to keep any large-scale background (e.g., scattered
360 # light from bright stars) from being selected for sky objects in
361 # the calculation, so do another detection pass without either the
362 # local or wide temporary background subtraction; the DETECTED
363 # pixels will mark the area to ignore.
364 originalMask = maskedImage.mask.array.copy()
365 try:
366 self.clearMask(exposure.mask)
367 convolveResults = self.convolveImage(maskedImage, psf, doSmooth=doSmooth)
368 tweakDetResults = self.applyThreshold(convolveResults.middle, maskedImage.getBBox(), factor)
369 self.finalizeFootprints(maskedImage.mask, tweakDetResults, sigma, factor=factor)
370 bgLevel = self.calculateThreshold(exposure, seed, sigma=sigma, minFractionSourcesFactor=0.5,
371 isBgTweak=True).additive
372 finally:
373 maskedImage.mask.array[:] = originalMask
374 self.tweakBackground(exposure, bgLevel, results.background)
376 return results
378 def tweakBackground(self, exposure, bgLevel, bgList=None):
379 """Modify the background by a constant value
381 Parameters
382 ----------
383 exposure : `lsst.afw.image.Exposure`
384 Exposure for which to tweak background.
385 bgLevel : `float`
386 Background level to remove
387 bgList : `lsst.afw.math.BackgroundList`, optional
388 List of backgrounds to append to.
390 Returns
391 -------
392 bg : `lsst.afw.math.BackgroundMI`
393 Constant background model.
394 """
395 self.log.info("Tweaking background by %f to match sky photometry", bgLevel)
396 exposure.image -= bgLevel
397 bgStats = lsst.afw.image.MaskedImageF(1, 1)
398 bgStats.set(bgLevel, 0, bgLevel)
399 bg = lsst.afw.math.BackgroundMI(exposure.getBBox(), bgStats)
400 bgData = (bg, lsst.afw.math.Interpolate.LINEAR, lsst.afw.math.REDUCE_INTERP_ORDER,
401 lsst.afw.math.ApproximateControl.UNKNOWN, 0, 0, False)
402 if bgList is not None:
403 bgList.append(bgData)
404 return bg
406 def _computeBrightDetectionMask(self, maskedImage, convolveResults):
407 """Perform an initial bright source detection pass.
409 Perform an initial bright object detection pass using a high detection
410 threshold. The footprints in this pass are grown significantly more
411 than is typical to account for wings around bright sources. The
412 negative polarity detections in this pass help in masking severely
413 over-subtracted regions.
415 A maximum fraction of masked pixel from this pass is ensured via
416 the config ``brightMaskFractionMax``. If the masked pixel fraction is
417 above this value, the detection thresholds here are increased by
418 ``bisectFactor`` in a while loop until the detected masked fraction
419 falls below this value.
421 Parameters
422 ----------
423 maskedImage : `lsst.afw.image.MaskedImage`
424 Masked image on which to run the detection.
425 convolveResults : `lsst.pipe.base.Struct`
426 The results of the self.convolveImage function with attributes:
428 ``middle``
429 Convolved image, without the edges
430 (`lsst.afw.image.MaskedImage`).
431 ``sigma``
432 Gaussian sigma used for the convolution (`float`).
434 Returns
435 -------
436 brightDetectedMask : `numpy.ndarray`
437 Boolean array representing the union of the bright detection pass
438 DETECTED and DETECTED_NEGATIVE masks.
439 """
440 # Initialize some parameters.
441 brightPosFactor = (
442 self.config.prelimThresholdFactor*self.config.brightMultiplier/self.config.bisectFactor
443 )
444 brightNegFactor = self.config.brightNegFactor/self.config.bisectFactor
445 nPix = 1
446 nPixDet = 1
447 nPixDetNeg = 1
448 brightMaskFractionMax = self.config.brightMaskFractionMax
450 # Loop until masked fraction is smaller than
451 # brightMaskFractionMax, increasing the thresholds by
452 # config.bisectFactor on each iteration (rarely necessary
453 # for current defaults).
454 while nPixDetNeg/nPix > brightMaskFractionMax or nPixDet/nPix > brightMaskFractionMax:
455 self.clearMask(maskedImage.mask)
456 brightPosFactor *= self.config.bisectFactor
457 brightNegFactor *= self.config.bisectFactor
458 prelimBright = self.applyThreshold(convolveResults.middle, maskedImage.getBBox(),
459 factor=brightPosFactor, factorNeg=brightNegFactor)
460 self.finalizeFootprints(
461 maskedImage.mask, prelimBright, convolveResults.sigma*self.config.brightGrowFactor,
462 factor=brightPosFactor, factorNeg=brightNegFactor
463 )
464 # Check that not too many pixels got masked.
465 nPix = maskedImage.mask.array.size
466 nPixDet = countMaskedPixels(maskedImage, "DETECTED")
467 self.log.info("Number (%) of bright DETECTED pix: {} ({:.1f}%)".
468 format(nPixDet, 100*nPixDet/nPix))
469 nPixDetNeg = countMaskedPixels(maskedImage, "DETECTED_NEGATIVE")
470 self.log.info("Number (%) of bright DETECTED_NEGATIVE pix: {} ({:.1f}%)".
471 format(nPixDetNeg, 100*nPixDetNeg/nPix))
472 if nPixDetNeg/nPix > brightMaskFractionMax or nPixDet/nPix > brightMaskFractionMax:
473 self.log.warn("Too high a fraction (%.1f > %.1f) of pixels were masked with current "
474 "\"bright\" detection round thresholds. Increasing by a factor of %f "
475 "and trying again.", max(nPixDetNeg, nPixDet)/nPix,
476 brightMaskFractionMax, self.config.bisectFactor)
478 # Save the mask planes from the "bright" detection round, then
479 # clear them before moving on to the "prelim" detection phase.
480 brightDetectedMask = (maskedImage.mask.array
481 & maskedImage.mask.getPlaneBitMask(["DETECTED", "DETECTED_NEGATIVE"]))
482 self.clearMask(maskedImage.mask)
483 return brightDetectedMask
486def countMaskedPixels(maskedIm, maskPlane):
487 """Count the number of pixels in a given mask plane.
489 Parameters
490 ----------
491 maskedIm : `lsst.afw.image.MaskedImage`
492 Masked image to examine.
493 maskPlane : `str`
494 Name of the mask plane to examine.
496 Returns
497 -------
498 nPixMasked : `int`
499 Number of pixels with ``maskPlane`` bit set.
500 """
501 maskBit = maskedIm.mask.getPlaneBitMask(maskPlane)
502 nPixMasked = np.sum(np.bitwise_and(maskedIm.mask.array, maskBit))/maskBit
503 return nPixMasked