Processing of exoplanet full field images

1
Processing of exoplanet full field images
Farid Karioty CoRoT Week 12/06/2005
2
Plan

• I. Already done
• Images corrections
• Stars identification
• MGPDV (CoRoT Sight Geometric Model) update
• PSF extraction from full field images
• II. To be done
• Masks assignment
• Background windows
• Offset windows
• Spectrum calculations for each star after mask
  assignment
• III. Remaining problem
• Photometric precision of extracted PSF
• IV. Conclusion

3
I. EXOWIND

• Inputs
• 3 full field images (equivalent exposition time
  is 30 min/ image)
• EXODAT extractions
• EXOBASKET
• Theoretic PSF set
• Outputs
• 2 positions files of these stars for MGPDV update
• XML assignment file of the masks

4
EXOWIND (IHM)
5
Cosmic impacts correction

• Method
• 3 full field images, 30 minutes exposition each
• For each pixel of the 3 images
• Calculate the median for each pixel triplet
• If a pixel value exceeds mean value by more than
  3s, then it is replaced by the median value of
  the 3 pixels
• The 3 images are summed

6
EMC correction (crosstalk)

• Crosstalk
• Depends of seismology channel windowing
• Scrambling on the exoplanet channel
• Correction
• Parasites positions are predictable
• Values of the different scrambling sequences are
  read in prescan pixels of the full field images
• An image containing the parasites is generated
  subtracted

7
Crosstalk correction(IHM)
8
Offset correction

• Method
• Calculate in prescan and overscan pixels the
  offset values for each half CCD
• Subtraction of the measured offset (possibility
  to choose between the value measured in the
  prescan or the overscan pixels)

9
Offset correction(IHM)
10
Gain correction

• Method
• Reading in the BDE (calibration data base) of the
  gain values for each channel
• Application of the multiplicative factor for each
  half CCD

11
Smearing correction

• Method
• Calculate the smearing value for each column of
  the image
• Smearing subtraction

12
Background correction

• Methods
• Division of the image into sub-images in which
  the minimum value is taken, then interpolation
  back to a 2048x2048 pixels image
• Same method but the median value is used
• Convolution method convolution of the image by
  an enlarged Gaussian fit by a 2nd degree
  polynomial
• Ravines search of valleys in the image

13
Background correction(IHM)
14
Identification of saturated stars

• Method
• Histogram of the image selection of the
  saturation threshold
• Research in the image of saturation domains
  (adjacent pixels with values greater to the
  chosen saturation threshold)
• Identification of these stars (automatic
  identification manual module for the stars
  where a doubt persists e.g. 2 close saturated
  stars)

15
Identification of saturated stars (IHM)
16
Stars identification

• Identification of about 20 bright slightly
  contaminated stars of the same spectral type
• Update of CCD position in the MGPDV (translation
  rotation)
• Identification of 100 to 500 stars (still of the
  same spectral type)
• Distortion update of the MGPDV

17
Stars identification (2)

• Method
• Projection of the catalogue on the CCD (selection
  of the stars with these 3 parameters mgr,
  contamination, spectral type)
• For each star calculation of the subpixel shift
  between the star position on the CCD and its
  position given by the catalogue the MGPDV
  (correlation method)
• If the shift is less than a user-defined limit
  (depending on the knowledge of CCD position
  distortion coefficients in the MGPDV) if the
  correlation is greater to a user-defined
  threshold, then the star is identified

18
Stars identification(IHM)
19
PSF extraction

• Method
• Selection in the catalogue of the stars
  corresponding to
• the PSF spectral type to extract
• the maximum magnitude of these PSF stars (MGR
  min MGR sat)
• the maximum contamination level of these stars
• Choice of the number of sub-domains in the image
  (1 extracted PSF by sub-domain and spectral type)
• Summation of the stack of PSF stars after
  subpixel recentering
• Filtering by an ellipsoidal Gaussian to decrease
  the background noise (residuals of the
  corrections, other fainter stars)

20
PSF extraction (IHM)
21
Extracted PSF
22
II. Masks assignment

• Method
• For each EXOBASKET star
• PSF fitting
• Fitted PSF signal, remaining noise
• Stack of images for the attribution procedure
• XML file of the masks assignments
• But it is crucial to know precisely the PSF

23
III. Unsolved problem

• Photometric precision of the extracted PSF
• Important remainders, maximum errors 20
• Too much important imprecision for a PSF fit
• assignment quality is decreased
• A deconvolution method is being implemented

24
IV. Conclusion

• Images corrections OK
• Identification of the saturated stars and of the
  saturation magnitude OK
• Identification of the stars OK
• PSF extraction not totally solved but the
  deconvolution method seems to give better results