LIBRARY ROUTINE
limedian - perform median filtering
LIBRARY CALL
int limedian(kobject in_obj, int wsize, int hsize, int reps, kobject out_obj)
INPUT
- in_obj - input object to be processed
- wsize - width of window in pixels
- hsize - height of window in pixels
- reps - number of repetitions of filter to perform
OUTPUT
RETURN VALUE
TRUE (1) on success, FALSE (0) otherwise
DESCRIPTION
for each WxH window in the image, replace the pixel
at (W/2)x(H/2) (integer arithmetic) with the median of
the values in the window. The median is computed by
sorting the values in the window with quicksort and
extracting the middle value as the median. This is
not the fastest approach for BYTE data where a
histogram updating method could be used. It is, however,
extensible to higher data types including floating
point data.
If a map is present, the data is mapped prior to
operation. Mask data is ignored.
This implementation processes the image data by
full planes.
Boundaries are handled by zero padding.
ADDITIONAL INFORMATION
none
EXAMPLES
none
SIDE EFFECTS
none
RESTRICTIONS
Complex data types are not handled at all.
MODIFICATION
none
FILES
$IMAGE/objects/library/kimage_proc/src/limedian.c
SEE ALSO
kimage_proc(3)
COPYRIGHT
Copyright (C) 1993 - 1997, Khoral Research, Inc. ("KRI") All rights reserved.