6. Very low level processing (radiometric calibration)

Size: px

Start display at page:

Download "6. Very low level processing (radiometric calibration)"

Abigayle Cooper
5 years ago
Views:

1 Master ISTI / PARI / IV Introduction to Astronomical Image Processing 6. Very low level processing (radiometric calibration) André Jalobeanu LSIIT / MIV / PASEO group Jan lsiit-miv.u-strasbg.fr/paseo PASEO

2 Very low level processing: radiometric calibration & corrections Sensor & instrument calibration Nonlinearity labeling/correction Artifact labeling/elimination (bad pixels & columns) Bias and dark correction Combined spatial response compensation (flat field) Wavelength calibration (multispectral) Sky effects removal Spatial corrections (cosmics, background, transparency) Spectral sky subtraction (sky emission, stray lights) Spectral calibration (atmospheric absorption)

3 Sensor/instrument radiometric calibration Know how to apply basic, pixelwise operations Find out how to use more complex techniques involving pixel neighbors Understand what processing is needed prior to simple corrections (calibration techniques)

Nonlinearity identification/correction Data-dependent effects CFHT Calibration transfer function saturation trails Measure pixel transfer

value Corrections Inverse transfer function - elementwise pixel transform Saturation: noninvertible transform Label saturated pixels to

4 Nonlinearity identification/correction Data-dependent effects CFHT Calibration transfer function saturation trails Measure pixel transfer function (intrinsic sensor nonlinearity) usually negligible Identify & label saturated regions thresholding, constant areas of max. value Corrections Inverse transfer function - elementwise pixel transform Saturation: noninvertible transform Label saturated pixels to avoid subsequent processing: missing data Attempt to predict the values: fill in the gaps - unknown high values... interpolation, Markov Random Field sampling

Correction Data loss: noninvertible transform Label bad pixels to avoid subsequent processing: missing

5 Artifact elimination Data-independent effects CFHT partially bad columns bad columns & regions Calibration Measure positions of bad pixels or bad columns (usually constant, but may worsen with time) Correction Data loss: noninvertible transform Label bad pixels to avoid subsequent processing: missing data Predict the values: fill in the gaps 1D or 2D interpolation, Markov Random Field sampling, rank filtering...

filtering to reduce noise Measure the dark current map (depends on temperature, negligible for cooled sensors) exposure

6 Bias and dark correction Data-independent, condition-dependent effects Calibration a dark frame a bias or offset frame Measure the bias map: additive bias for each pixel (relatively constant over time) very short exposures, image sequence filtering to reduce noise Measure the dark current map (depends on temperature, negligible for cooled sensors) exposure time close to integration time, denoising may be needed Corrections Elementwise bias and dark subtraction Corrected0 = Obs-Dark-Bias

flat-field Combined sensor, optical & defects Multiple exposures (flat source), image sequence filtering to

7 Combined spatial response compensation C. Buil flat-field raw, observed image corrected image Calibration Measure the combined spatial sensitivity map: flat-field Combined sensor, optical & defects Multiple exposures (flat source), image sequence filtering to reduce noise Don t forget the dark and bias correction! Correction Elementwise division by flat-field map Corrected1=Corrected0/Flat Corrected1=(Obs-Dark-Bias)/(ObsFlat-DarkFlat-Bias)

Wavelength calibration (multispectral) Preprocessing Dark current and bias removal Spatial effects removal Calibration - multispectral imaging Spectral sensitivity measurement for each band Neon arc

8 Wavelength calibration (multispectral) Preprocessing Dark current and bias removal Spatial effects removal Calibration - multispectral imaging Spectral sensitivity measurement for each band Neon arc calibration for SAURON IFS Overall instrument sensitivity: use calibrated black body sources Calibration - integral field spectroscopy Wavelength position estimation on the sensor Use calibrated sources (emission lines) to build the wavelength map Spectral sensitivity measurement Overall instrument sensitivity: use calibrated black body sources Corrections Multispectral imaging: simple division Global image division by sensitivity factor Integral field spectroscopy: division by spectral response Elementwise division by sensitivity map = sensitivity(wavelength map)

9 Sky-related radiometry compensation Find out how to remove random sky effects on single band images Understand the basics of skyrelated spectral calibration Anthony Arrigo

morphology Look for linear trails Hough transform Correction Data loss: noninvertible degradation Label cosmic rays:

10 Cosmic ray labeling/removal Random process Detection Look for bright, saturated, sharp objects Thresholding using neighborhood information, math. morphology Look for linear trails Hough transform Correction Data loss: noninvertible degradation Label cosmic rays: grazing incidence (left), normal (right) bad pixels to avoid subsequent processing: missing data Predict the values: fill in the gaps 1D or 2D interpolation, Markov Random Field sampling, rank filtering...

Background and transparency corrections spatial effects, wide FOV Random, condition-dependent effects Background calibration Assumption: black background Simple, uniform

), linear regression Transparency calibration Simple, uniform transparency over the FOV Standard star photometry, average Linear or low order polynomial model Standard

11 Background and transparency corrections spatial effects, wide FOV Random, condition-dependent effects Background calibration Assumption: black background Simple, uniform background over the FOV Background selection (thresh.), average Linear or low order polynomial model Background selection (thresh.), linear regression Transparency calibration Simple, uniform transparency over the FOV Standard star photometry, average Linear or low order polynomial model Standard star photometry, linear regression Corrections Background: model subtraction Transparency: division by transparency model Combined: Corrected3=(Corrected1-SkyBack)/SkyTrans C. Buil Background fitting & subtraction

background Find lines, fit continuum ( sky fibers in fiber-fed IFS: very narrow FOV) Correction Subtraction Elementwise

12 Spectral calibration: sky background IFS, narrow FOV Vertical sky emission lines (NIFS instrument) Example of sky emission measurement Calibration Estimate sky continuum & emission lines Aim at black areas (no stars or objects): spectrum of the background Find lines, fit continuum ( sky fibers in fiber-fed IFS: very narrow FOV) Correction Subtraction Elementwise subtraction of the sky emission model or measured spectrum Also applicable to multispectral images for a better spectral accuracy

13 Spectral calibration: atmospheric opacity IFS, narrow FOV Atmospheric transmission at Mauna Kea (ULTRAM model, I. Chapman) Calibration Use atmospheric opacity models Calibrate the models: measure temperature, pressure, humidity Use reference spectra Standard stars or objects databases (e.g. observed from space) Laser beam (lidar): measure various parameters Correction Division by wavelength-dependent opacity Elementwise division by the atmospheric opacity model Also applicable to multispectral images for a better spectral accuracy

14 Radiometric correction examples CCD data reduction (HST) Flow of Spectroscopic Data through calstis-6, 1D Extraction 2D CCD Data Reduction Common to Imaging and Spectroscopy

a simple optical imager

Imagers and Imaging a simple optical imager Here s one on our 61-Inch Telescope Here s one on our 61-Inch Telescope filter wheel in here dewar preamplifier However, to get a large field we cannot afford