5.1 DrizzlePac: An Overview

DrizzlePac is a package of tasks used primarily for registration and resampling of images after the calibration pipeline. AstroDrizzle and TweakReg are its flagship tasks, and the rest of DrizzlePac’s tasks support other drizzling operations.

AstroDrizzle uses the same historic algorithmic base implemented in previous drizzling software with a modified set of core routines that have been recoded in C and Python. One notable change in comparison to previous versions of drizzling software is that AstroDrizzle moves pixels according to the astrometry encoded in the WCS, as opposed to doing so according to shifts. 

This chapter provides a description of the software interface to AstroDrizzle and the other tasks in the DrizzlePac package.

In a nutshell, AstroDrizzle processes a set of pipeline flat-field calibrated files (flc.flt/flt.fits) as follows:

• Using astrometric information and distortion information found in the header, the images are separately drizzled onto undistorted output images in a common reference frame.
• These distortion-free output images are combined to create a median image. The median image is the first approximation of the "truth" image. 
• The median image is then blotted, or reverse-drizzled, back to the frame of each input flc.flt/flt.fits image.
• By comparing each flc.flt/flt.fits image with its blotted counterpart median image, the software locates bad pixels in each of the original flc.flt/flt.fits frames and creates bad pixel masks. These bad pixels are typically cosmic rays or faulty pixels in the detector.
• In the final step, all images are drizzled, using the mask files, onto a single output image. This produces an image that’s corrected for geometric distortion and largely free of cosmic rays and detector artifacts.

In the pipeline, AstroDrizzle task parameters are given default values selected to cover a wide range of data. These parameter values are stored in a reference file named by the MDRIZTAB image header keyword. However, these values may not be well suited for some images. In such cases, users may elect to reprocess the images to obtain better quality results.

In this section, information will be provided to help users select the best possible task parameter values suited for reprocessing their data. This includes combining images spread over several visits at different roll angles–this type of data is almost always taken using different guide stars, and due to positional uncertainties in the Guide Star Catalog, the coordinate frames of each visit could be significantly misaligned as much as 0.5 arcseconds. A DrizzlePac task called tweakreg will serve as a useful tool, in most imaging cases, for aligning all multi-visit images to a common WCS. There are also tools for converting pixel positions to sky coordinates, and vice versa, as well as translating pixel positions for an image before and after geometric distortion corrections.