compare -- align the surface of images using a polynomial fit

`compare [flags] image(s)`

kappa | k | 4.0 |

nan | n | 0.0 |

order | i | 2 |

radius | l | 10 |

reference | f | |

region | r | 0:0,0:0 |

verbose | v | 0 |

- kappa = 4.0
- Cutting parameter. Number of sigmas for clipping outlying pixels (caused e.g. by stars).

- nan = 0.0
- Not a number value. Pixel with this value are not taken into account and stay at "nan" in the output image.

- order = 2
- Order of polynomial fit:
- 0: const.
- 1: const. + a*x
^{1}+ b*y^{1} - 2: const. + a*x
^{1}+ b*y^{1}+ c*x^{2}+ d*x^{1}y^{1}+ e*y^{2} - ...

- radius = 10
- Cutting radius. If a pixel's value p(x,y) > min_3 + kappa * e(x,y) or if it's value p(x,y) < max_3 - kappa * e(x,y) the pixel is not taken into account in the step of the polynomial surface fit. min_3 is the the third minimum, max_3 the third maximum value of all neighbouring pixels closer than the cutting radius and e(x,y) is the pixel's error value (measured in the errorframe).

- reference
- Image to which the comparison is made. If no image is specified all
`image(s)`are compared to the first.

- region = 0:0,0:0
- Normalization Region. The median of this region stays the same for the input image and the compared output image. If region is equal to "0:0,0:0" the region will be a centered rectangle with a quarter of the image's lengths.

- verbose = 0
- Verbose level. If
`verbose`>= 1, for each input image three additional output images are written showing the cutted pixels and the fitted polynomial surface.

- image(s)
- Image(s) to be processed.

The task compare is designed for aligning the surface (e.g. the illumination of a flatfield) of one or more `image(s)` to the surface of a
`reference` image. The alignment is done by multiplication with a two-dimentional surface of a given `order`. To determine this surface
the task first divides the reference image by the single image and normalizes the result to the median of all pixels in a given `region` (in
order to have roughly the same flux before and after the alignment). After that remaining stars originating from either the reference image or the
single image are cutted: within a `radius` it is checked if a pixel's value p(x,y) > min_3 + `kappa` * e(x,y) or if it's value p(x,y)
< max_3 - `kappa` * e(x,y). min_3 is the the third lowest, max_3 the third highest value of all neighbouring pixels closer than the cutting
`radius` and e(x,y) is the pixel's error value (measured in the errorframe). Using all the remaining pixels the two-dimentional surface is
determined and finaly multiplied with the single image.

If `verbose` >= 1, for each input image three additional output images are written showing the cutted pixels and the fitted polynomial
surface.

- c...
- aligned images

- e.c...
- errorframes of aligned images

- If
`verbose`>= 1:

- div-...
- This output images are the result of a division of the reference image by each of the input images. A border of 10 pixels is added.

- cutdiv-...
- Same as div-... but all the cutted pixels are set to the value "0".

- fit-...
- Polynomial surface fits of the normalized division of the reference image by each of the input images. The input images are multiplied by this in order to get the aligned images.

compare -i 2 -k 4 -f b-sky_r_040529_001.fits b-sky_r_040529_*.fits

Dokumentation maintained by J. Koppenhöfer

This file was last time updated on 21 Feb 2005