method used employs a rolling histogram and allows the whole image to be filtered.
The median pixel value in the region surrounding each input image pixel is calculated. This value is
then subtracted from the value of the pixel being considered. Finally, the original background count is
added to the result, which is placed in the corresponding pixel of the output image.
FASTMED IN OUT BACK SIGMA WIDTH
BACK = _REAL (Read)
background count value. Units counts.
IN = _NDF (Read)
The name of the NDF to which the
filter will be applied.
OUT = _NDF (Write)
The name of the output NDF that will be
SIGMA = _REAL (Read)
Standard deviation of the background count value.
WIDTH = _INTEGER (Read)
The width of the filter to be employed. Units
fastmed in=field out=flatgal back=760. sigma=27. width=72 In this
example, a 72x72 pixel filter will be applied to the input image FIELD. The resulting median
filtered image will be placed in output image FLATGAL. The input image (FIELD) had a
global background count value of 760 with an associated standard deviation of 27 counts.
With small filters it may be found that the resulting output images are noisy.
This is due to the small number of pixels contributing to the histogram. The problem will
be most obvious at the image edges and corners. For this reason some users may find it
necessary to clip the output image by WIDTH/2 pixels on each edge to generate better results.