Resamples and mosaics using the drizzling algorithm DRIZZLE
The algorithm used for combining the images on the output grid is Variable-Pixel Linear
Reconstruction, or so-called ‘
drizzling’
. The user is allowed to shrink the input pixels to a smaller
size (drops) so that each pixel of the input image only affects pixels in the output image under the
corresponding drop.
If the logging system has been initialised using CCDSETUP, then the value specified there will be
used. Otherwise, the default is "
CCDPACK.LOG"
. [CCDPACK.LOG]
TERMINAL – Send output to the terminal only
LOGFILE – Send output to the logfile only (see the LOGFILE parameter)
BOTH – Send output to both the terminal and the logfile
NEITHER – Produce no output at all
If the logging system has been initialised using CCDSETUP, then the value specified there will be
used. Otherwise, the default is "
BOTH"
. [BOTH]
If weighting of the input pixels by the mean inverse variance of the entire input image (rather than the pixels own variance) is required MAPVAR should be set to .FALSE. and USEVAR should be set to .TRUE. (this is the default condition). [FALSE]
"
drop"
size, this being the ratio of the linear size of the drizzled
drop to that of the input pixel. Interlacing is equivalent to setting PIXFRAC=0.0, while
shift-and-add is equivalent to setting PIXFRAC=1.0. For low values of PIXFRAC the MULTI
parameter must also be set correspondingly low. [0.9] "
preserved"
in the output image. Alternatively, if a FALSE
value is given, then the output image will be given an appropriate floating point data
type.
When using integer input data, the former option is useful for minimising the storage space required for large mosaics, while the latter typically permits a wider output dynamic range when necessary. A wide dynamic range is particularly important if a large range of scale factor corrections are being applied (as when combining images with a wide range of exposure times).
If a global value has been set up for this parameter using CCDSETUP, then that value will be used. [TRUE]
"
reference image"
supplied via this
parameter.
If scale-factor, zero-point corrections (see the SCALE and ZERO parameters respectively) have not
been specified via a sequential file listing (see the CORRECT parameter) then if an image is given via
the REF parameter the program will attempt to normalise the input images to the "
reference image"
supplied.
This provides a means of retaining the calibration of a set of data, even when corrections are being applied, by nominating a reference image which is to remain unchanged. It also allows the output mosaic to be normalised to any externally-calibrated image with which it overlaps, and hence allows a calibration to be transferred from one set of data to another.
If the image supplied via the REF parameter is one of those supplied as input via the IN parameter, then this serves to identify which of the input images should be used as a reference, to which the others will be adjusted. In this case, the scale-factor, zero-point corrections and/or weightings applied to the nominated input image will be set to one, zero and one respectively, and the corrections for the others will be adjusted accordingly.
Alternatively, if the reference image does not appear as one of the input images, then it will be included as an additional set of data in the inter-comparisons made between overlapping images and will be used to normalise the corrections obtained (so that the output mosaic is normalised to it). However, it will not itself contribute to the output mosaic in this case. [!]
If SCALE is set to TRUE, then DRIZZLE will ask the user for a sequential file containing the corrections for each image (see the CORRECT parameter). If none is supplied the program will attempt to find its own corrections.
DRIZZLE will inter-compare the images supplied as input and will estimate the relative scale-factor between selected pairs of input data arrays where they overlap. From this information, a global set of multiplicative corrections will be derived which make the input data as mutually consistent as possible. These corrections will be applied to the input data before drizzling them onto the output frame.
Calculation of scale-factor corrections may also be combined with the use of zero-point corrections (see the ZERO parameter). By default, no scale-factor corrections are applied. [FALSE]
If weighting of the input image by the inverse variance map (rather than the mean) then the MAPVAR parameter whould be used. [TRUE]
If ZERO is set to TRUE, then DRIZZLE will ask the user for a sequential file containing the corrections for each image (see the CORRECT parameter). If none is supplied the program will attempt to calculate its own corrections.
DRIZZLE will inter-compare the images supplied as input and will estimate the relative zero-point difference between selected pairs of input data arrays where they overlap. From this information, a global set of additive corrections will be derived which make the input data as mutually consistent as possible. These corrections will be applied to the input data before drizzling them onto the output frame.
Calculation of zero-point corrections may also be combined with the use of scale-factor corrections (see the SCALE parameter). By default, no zero-point corrections are applied. [FALSE]
"
"
into a mosaic called "
out"
. The drop size of the input pixel is set to 0.7, i.e. it is scaled
to 70% of its orginal size before being drizzled onto the output grid. "
img"
into a mosaic called "
combined"
. Both scaling and zero-point corrections are enabled (the program
will request a correction file), however no reference image has been supplied (the program will use the
first image supplied in the input list). The multiplicative scaling factor between input and output
images is set to 4, i.e. the input pixel is 4 times larger than the output pixel and contains 16 output
pixels. INDEX SCALE ZERO
Where the fields have the following meaning:
"
#"
character. "
A package for the reduction of dithered
undersampled images"
, in Casertano et al. (eds), HST Calibration Workshop, STSCI, 1997, pp.
518–528:
The drizzle algorithm is conceptually straightforward. Pixels in the original
input images are mapped into pixels in the subsampled output image, taking into
account shifts and rotations between the images and the optical distortion of the
camera. However, in order to avoid convolving the image with the larger pixel
‘
footprint’
of the camera, we allow the user to shrink the pixel before it is averaged
into the output image.
The new shrunken pixels, or ‘
drops’
, rain down upon the subsampled output.
In the case of the Hubble Deep Field (HDF), the drops used had linear dimensions
one-half that of the input pixel – slightly larger than the dimensions of the output
subsampled pixels. The value of an input pixel is averaged into the output pixel
with a weight proportional to the area of overlap between the ‘
drop’
and the
output pixel. Note that, if the drop size if sufficently small, not all output pixels
have data added to them from each input image. One must therefore choose a drop
size that is small enough to avoid degrading the image, but large enough so that
after all images are ‘
dripped’
the coverage is fairly uniform.
The drop pize if controlled by a user-adjustable parameter called PIXFRAC, which is simply the ratio of the linear size of the drop to the input pixel (before any adjustment due to geometric distortion of the camera). Thus interlacing is equivalent to setting PIXFRAC=0.0, while shift-and-add is equivalent to PIXFRAC=1.0.
When a drop with value and a user-defined weight is added to an image with pixel value , weight , and fractional pixel overlap , the resulting value the image and weight is
This algorithm has a number of advantages over standard linear reconstruction methods presently used. Since the area of the pixels scales with the Jacobian of the geometric distortion, drizzle preserves both surface and absolute photometry. Therefore flux can be measured using an aperture whose size is independent of position on the chip. As the method anticipates that a given output pixel may receive no information from a given input pixel, missing data (due for instance to cosmic rays or detector defects) do not cause a substantial problem, so long as there are enough dithered images to fill in the gaps caused by these zero-weight input pixels. Finally the linear weighting scheme is statistically optimum when inverse variance maps are used as weights.
All non-complex numeric data types are supported.
Bad pixels are supported.
The algorithm is restricted to handling 2D images only