70 likes | 73 Views
Learn about the importance of masks, weights, and flatfields in image processing. Explore different types of masks and their applications, such as for image statistics, sextractor, coaddition, and cosmic ray removal. Discover how to create weight frames and error maps for better error propagation.
E N D
Masks, weights and flatfields Roeland Masks and weights Philippe Are WFI flatfields flat? Ewout Making a supersky?
Masks, for what • Image statistics • We could do this more often • Making weights • weight = flat * mask • for sextractor • for coaddition • Other, • cosmic ray program
Different Masks • Cold pixels • bias > mean(bias) + 5*rms(bias) • Hot pixels • flat / convolved(flat) < 0.95 • Cosmic rays • Satellite tracks • Saturated pixels • ‘low’ gain (co-addition only)
Current situation class ColdPixelFrame class HotPixelFrame class MaskFrame mask = coldpixels & hotpixels class MasterWeight weight = master_flat * mask_frame class FlagFrame mask = thresh(weight) * science.cosmic() class WeightFrame weight = master_weight * flag.as_mask()
New situation • Add all operations • Use eclipse.pixelmap object • 1=good, 0=bad • (de)compress on-the-fly • clean up class-hierarchy mess • simplify ScienceFrame.make()
New Situation class ColdPixelMap class HotPixelMap class CosmicMap class SatelliteMap ? class Weight weight = flat * cold&hot&cosmic&sat weight = weight * thresh(science) weight = weight * thresh(weight)
Error maps • Weight is ‘relative error map’ • Could make absolute error map and do proper propagation of errors • Nice idea • but, remember the applications