CrosstalkCalib

class lsst.ip.isr.CrosstalkCalib(detector=None, nAmp=0, **kwargs)

Bases: IsrCalib

Calibration of amp-to-amp crosstalk coefficients.

Parameters

detectorlsst.afw.cameraGeom.Detector, optional

Detector to use to pull coefficients from.

nAmpint, optional

Number of amplifiers to initialize.

loglogging.Logger, optional

Log to write messages to.

**kwargs :

Parameters to pass to parent constructor.

Notes

The crosstalk attributes stored are:

hasCrosstalkbool

Whether there is crosstalk defined for this detector.

nAmpint

Number of amplifiers in this detector.

crosstalkShapetuple [int, int]

A tuple containing the shape of the coeffs matrix. This should be equivalent to (nAmp, nAmp).

coeffsnumpy.ndarray

A matrix containing the crosstalk coefficients. coeff[i][j] contains the coefficients to calculate the contribution amplifier_j has on amplifier_i (each row[i] contains the corrections for detector_i).

coeffErrnumpy.ndarray, optional

A matrix (as defined by coeffs) containing the standard distribution of the crosstalk measurements.

coeffNumnumpy.ndarray, optional

A matrix containing the number of pixel pairs used to measure the coeffs and coeffErr.

coeffValidnumpy.ndarray, optional

A matrix of Boolean values indicating if the coefficient is valid, defined as abs(coeff) > coeffErr / sqrt(coeffNum).

coeffsSqrnumpy.ndarray, optional

A matrix containing potential quadratic crosstalk coefficients (see e.g., Snyder+21, 2001.03223). coeffsSqr[i][j] contains the coefficients to calculate the contribution amplifier_j has on amplifier_i (each row[i] contains the corrections for detector_i).

coeffErrSqrnumpy.ndarray, optional

A matrix (as defined by coeffsSqr) containing the standard distribution of the quadratic term of the crosstalk measurements.

interChipdict [numpy.ndarray]

A dictionary keyed by detectorName containing coeffs matrices used to correct for inter-chip crosstalk with a source on the detector indicated.

Version 1.1 adds quadratic coefficients, a matrix with the ratios of amplifiers gains per detector, and a field to indicate the units of the numerator and denominator of the source and target signals, with “adu” meaning “ADU / ADU” and “electron” meaning “e- / e-“.

Version 1.2 adds the original gains used in the crosstalk fit.

Methods Summary

calculateBackground(mi[, badPixels])	Estimate median background in image.
extractAmp(image, ampToFlip, ampTarget[, ...])	Extract the image data from an amp, flipped to match ampTarget.
fromDetector(detector[, coeffVector, ...])	Set calibration parameters from the detector.
fromDict(dictionary)	Construct a calibration from a dictionary of properties.
fromTable(tableList)	Construct calibration from a list of tables.
subtractCrosstalk(thisExposure[, ...])	Subtract the crosstalk from thisExposure, optionally using a different source.
toDict()	Return a dictionary containing the calibration properties.
toTable()	Construct a list of tables containing the information in this calibration.
updateMetadata([setDate])	Update calibration metadata.

Methods Documentation

static calculateBackground(mi, badPixels=['BAD'])

Estimate median background in image.

Getting a great background model isn’t important for crosstalk correction, since the crosstalk is at a low level. The median should be sufficient.

Parameters

milsst.afw.image.MaskedImage

MaskedImage for which to measure background.

badPixelslist of str

Mask planes to ignore.

Returns

bgfloat

Median background level.

static extractAmp(image, ampToFlip, ampTarget, isTrimmed=False, fullAmplifier=False, parallelOverscan=None)

Extract the image data from an amp, flipped to match ampTarget.

Parameters

imagelsst.afw.image.Image or lsst.afw.image.MaskedImage

Image containing the amplifier of interest.

amplsst.afw.cameraGeom.Amplifier

Amplifier on image to extract.

ampTargetlsst.afw.cameraGeom.Amplifier

Target amplifier that the extracted image will be flipped to match.

isTrimmedbool, optional

The image is already trimmed.

fullAmplifierbool, optional

Use full amplifier and not just imaging region.

parallelOverscanbool, optional

This has been deprecated and is unused, and will be removed after v29.

Returns

outputlsst.afw.image.Image

Amplifier from image, flipped to desired configuration. This will always return a copy of the original data.

fromDetector(detector, coeffVector=None, coeffSqrVector=None)

Set calibration parameters from the detector.

Parameters

detectorlsst.afw.cameraGeom.Detector

Detector to use to set parameters from.

coeffVectornumpy.array, optional

Use the detector geometry (bounding boxes and flip information), but use coeffVector instead of the output of detector.getCrosstalk().

coeffSqrVectornumpy.array, optional

Quadratic crosstalk coefficients.

Returns

caliblsst.ip.isr.CrosstalkCalib

The calibration constructed from the detector.

classmethod fromDict(dictionary)

Construct a calibration from a dictionary of properties.

Must be implemented by the specific calibration subclasses.

Parameters

dictionarydict

Dictionary of properties.

Returns

caliblsst.ip.isr.CalibType

Constructed calibration.

Raises

RuntimeError

Raised if the supplied dictionary is for a different calibration.

classmethod fromTable(tableList)

Construct calibration from a list of tables.

This method uses the fromDict method to create the calibration, after constructing an appropriate dictionary from the input tables.

Parameters

tableListlist [lsst.afw.table.Table]

List of tables to use to construct the crosstalk calibration.

Returns

caliblsst.ip.isr.CrosstalkCalib

The calibration defined in the tables.

subtractCrosstalk(thisExposure, sourceExposure=None, crosstalkCoeffs=None, crosstalkCoeffsSqr=None, crosstalkCoeffsValid=None, badPixels=['BAD'], minPixelToMask=45000, doSubtrahendMasking=False, crosstalkStr='CROSSTALK', isTrimmed=None, backgroundMethod='None', doSqrCrosstalk=False, fullAmplifier=False, parallelOverscan=None, detectorConfig=None, badAmpDict=None, ignoreVariance=False)

Subtract the crosstalk from thisExposure, optionally using a different source.

We set the mask plane indicated by crosstalkStr in a target amplifier for pixels in a source amplifier that exceed minPixelToMask, if doSubtrahendMasking is False. With that enabled, the mask is only set if the absolute value of the correction applied exceeds minPixelToMask. Note that the correction is applied to all pixels in the amplifier, but only those that have a substantial crosstalk are masked with crosstalkStr.

The uncorrected image is used as a template for correction. This is good enough if the crosstalk is small (e.g., coefficients < ~ 1e-3), but if it’s larger you may want to iterate.

Parameters

thisExposurelsst.afw.image.Exposure

Exposure for which to subtract crosstalk.

sourceExposurelsst.afw.image.Exposure, optional

Exposure to use as the source of the crosstalk. If not set, thisExposure is used as the source (intra-detector crosstalk).

crosstalkCoeffsnumpy.ndarray, optional.

Coefficients to use to correct crosstalk.

crosstalkCoeffsSqrnumpy.ndarray, optional.

Quadratic coefficients to use to correct crosstalk.

crosstalkCoeffsValidnumpy.ndarray, optional

Boolean array that is True where coefficients are valid.

badPixelslist of str, optional

Mask planes to ignore.

minPixelToMaskfloat, optional

Minimum pixel value to set the crosstalkStr mask plane. If doSubtrahendMasking is True, this is calculated from the absolute magnitude of the subtrahend image. Otherwise, this sets the minimum source value to use to set that mask.

doSubtrahendMaskingbool, optional

If true, the mask is calculated from the properties of the subtrahend image, not from the brightness of the source pixel.

crosstalkStrstr, optional

Mask plane name for pixels greatly modified by crosstalk (above minPixelToMask).

isTrimmedbool, optional

This option has been deprecated and does not do anything. It will be removed after v29.

backgroundMethodstr, optional

Method used to subtract the background. “AMP” uses amplifier-by-amplifier background levels, “DETECTOR” uses full exposure/maskedImage levels. Any other value results in no background subtraction.

doSqrCrosstalk: bool, optional

Should the quadratic crosstalk coefficients be used for the crosstalk correction?

fullAmplifierbool, optional

Use full amplifier and not just imaging region.

parallelOverscanbool, optional

This option is deprecated and will be removed after v29.

detectorConfiglsst.ip.isr.overscanDetectorConfig, optional

Per-amplifier configs to use if parallelOverscan is True. This option is deprecated and will be removed after v29.

badAmpDictdict [str, bool], optional

Dictionary to identify bad amplifiers that should not be source or target for crosstalk correction.

ignoreVariancebool, optional

Ignore the variance plane when doing crosstalk correction?

Notes

For a given image I, we want to find the crosstalk subtrahend image CT, such that

I_corrected = I - CT

The subtrahend image is the sum of all crosstalk contributions that appear in I, so we can build it up by amplifier. Each amplifier A in image I sees the contributions from all other amplifiers B_v != A. For the current linear model, we set sImage equal to the segment of the subtrahend image CT corresponding to amplifier A, and then build it up as: simage_linear = sum_v coeffsA_v * (B_v - bkg_v) where coeffsA_v is the vector of crosstalk coefficients for sources that cause images in amplifier A. The bkg_v term in this equation is identically 0.0 for all cameras except obs_subaru (and is only non-zero there for historical reasons). To include the non-linear term, we can again add to the subtrahend image using the same loop, as:

simage_nonlinear = sum_v (coeffsA_v * B_v) + (NLcoeffsA_v * B_v * B_v)

= sum_v linear_term_v + nonlinear_term_v

where coeffsA_v is the linear term, and NLcoeffsA_v are the quadratic component. For LSSTCam, it has been observed that the linear_term_v >> nonlinear_term_v.

toDict()

Return a dictionary containing the calibration properties.

The dictionary should be able to be round-tripped through fromDict.

Returns

dictionarydict

Dictionary of properties.

toTable()

Construct a list of tables containing the information in this calibration.

The list of tables should create an identical calibration after being passed to this class’s fromTable method.

Returns

tableListlist [lsst.afw.table.Table]

List of tables containing the crosstalk calibration information.

updateMetadata(setDate=False, **kwargs)

Update calibration metadata.

This calls the base class’s method after ensuring the required calibration keywords will be saved.

Parameters

setDatebool, optional

Update the CALIBDATE fields in the metadata to the current time. Defaults to False.

kwargs :

Other keyword parameters to set in the metadata.