lytle@noao.UUCP (04/08/87)
I am looking for information about fast algorithms for removing binary
wrap-around from two-dimensional data arrays. An example of the problem
I wish to solve is --- data stored in an image with 8 bits per pixel where
the dynamic range of the data is, say, 11 bits. If the data varies
somewhat smoothly from small values to large values the wrap-around will
make the image look like a contour map.
I want to unwrap this data into an image with enough range to hold the
unwraped data. I have invented a couple algorithms for this, the simplest
of which is to go through the image line by line and search each line
for pixel to pixel differences larger than some predetermined threshold,
keep track of the number of wraps and then add the appropriate value to
each pixel to unwrap it.
I have looked at more complicated algorithms used in phase unwrapping
in the optical sciences, cepstrum analysis, cellular automata, etc.
but most of these approaches are much too time consuming.
Any information and/or algorithm information would be greatly
appreciated.
Dyer Lytle
--
Dyer Lytle, National Optical Astronomy Observatories, Tucson, AZ, 602-323-4136
{arizona,decvax,hao,ihnp4,seismo}!noao!lytle noao!lytle@lbl-csam.ARPA