The filtering is done by two specially developed Fortran programmes. The the detection phase, for each pixel, the same pixel (position) in all comparison images is examined. If the pixel value deviation is more than KAPPA times the noise (KAPPA is often set to 2) the pixel is regarded as deviant. A mask image is continously updated and set to 1 where cosmic ray events are detected. This mask image is "grown" by 2 pixels in all directions. In the filtering phase, only positions marked in the mask image are filtered. Deviant pixels are replaced with a median of the comparison images.
Note that filtering is time consuming. About 10 mins CPU time per image can be expected, thus a total of about 2 hours CPU time.
|
21 |
@@ mcatal pc | M | Create catalogue for reference images |
| 22 | mfilter_script yframes y | U | Create filtering scripts |
| 23 | del/disp | M | Free the Midas image display |
| 24 | @@ filt >filt.log | M | Execute filtering |
Check your filtered images by comparing them to the unfiltered versions.