slosin Documentation

slosin - Remove periodic noise from LEVEL1 images

SLOSIN removes periodic noise, in the sample direction, from VIKING
images.  The images should be run through LEVEL1 before SLOSIN is
run on the images.

SLOSIN performs the following steps when processing an image:

1) A working window is defined by the user.  This window is
   defined by the STAL, STAS, ENDL, and ENDS parameters.  The
   window is where SLOSIN gets its information to determine the
   wavelength of the noise in the image.

2) A High Pass Filter is performed on the working window.  This
   extracts the image data and leaves only the noise behind.
   The size of the filter is 1 X FILT, where FILT is a TAE parameter.

3) A Low Pass Filter is now performed on the working window.
   This smooths the data, again trying to remove any image data
   that is left from the High Pass Filter.  The size of the
   filter is 1 X (FILT/2), where again FILT is a TAE parameter.

4) A Fourier transform is now applied on the working window to
   determine the wavelength of the noise.

5) If noise is found, then both the High and Low pass filters
   are run again on the working window.  The filter size is
   now defined as 1 X NOISE and 1 X (NOISE/2) respectively.
   The additional filtering reduces the chances of image data not
   being removed from the working window.  A model of the noise
   is constructed by adding all of the pixels, from the working
   window, to the proper locations along the wavelength.  The
   model is then divided by the pixel count at each position,
   returning an average noise value.  This noise model is then
   subtracted from the entire image line, not just the sub-line
   in the working window.

6) If the WAVE parameter was not assigned a value and noise was
   just removed, then the program resumes at step two and restarts
   the process.  The noise removal process continues for a max of
   two times or until SLOSIN does not detect anymore noise in the
   window.

7) The next user defined window is used as the working window
   and the program continues back at step 2.

                                HINTS
SLOSIN tries to remove the original image and leave only the noise
in the working window.  If the working window is defined over a high
contrast area, such as the polar cap or a planet limb, the filters
remove both image and noise from the working window, and SLOSIN is
not able to remove the noise.  A solution to this problem is to
define windows that do NOT include relatively high contrast data.
Another problem occurs when the working window that is defined is too
small.  To build a good model of the noise, it is desirable that SLOSIN
gather statistics of the noise over several noise cycles.  Several
noise cycles help average out the remaining image data.  If
too few cycles are used, a data glitch will offset the noise
model and will cause entire line to be "corrected" with this bad
model.

SLOSIN has a hard time finding noise wavelengths that are relatively
small (i.e. around 41).  There are two possible solutions.  One is
to get to know the wavelengths that are relatively small.  The two
most common small wavelengths are either 41 or 43.  The second solution
is to adjust the FILT parameter.  This parameter is used as a first
guess for the max noise wavelength.  If this parameter is too  small
or too large it will cause both the High pass and Low pass filters
to wipe out the noise from the working window.

                              LIMITATIONS
SLOSIN works on both 8 and 16 bit data.  However due to integer
round off, the 16-bit image data format is suggested.  Also, the
max value that FILT can contain is 512.

PROGRAMMER: Kim T. Thompson 2 Sep 97

INPUT IMAGE FILE NAME

NONE

OUTPUT IMAGE FILE NAME

NONE

STARTING LINE OF WINDOW

1

ENDING LINE OF WINDOW

10

STARTING SAMPLE OF WINDOW

30,30,30,30

ENDING SAMPLE OF WINDOW

1170,1170,1170,1170

WAVE LENGTH OF THE NOISE
( 0 = CALCULATE IT )

0

ESTITMATED WAVE LENGTH
(USED ONLY WHEN WAVE=0)

150

ADDITIONAL NOTES:

Parm	Description
FROM	Input image file name. This image has been through LEVEL1 processing.
TO	Output image file name. This output image has had SLOSIN processing performed on it.
STAL	This array contains all of the starting lines of the windows. These values have to be larger than 0 and less than the ending image line number. Also, these values must be LESS than those corresponding ENDL values.
ENDL	This array contains all of the ending lines of the windows. These values have to be larger than 0 and less than the ending image line number. Also, these values must be LARGER than those corresponding STAL values.
STAS	This array contains all of the starting samples of the windows. These values have to be larger than 0 and less than the ending image sample number. Also, these values must be LESS than those corresponding ENDS values. Defaults to 30
ENDS	This array contains all of the ending samples of the windows. These values have to be larger than 0 and less than the ending image sample number. Also, these values must be LARGER than those corresponding STAS values. Defaults to 1170
WAVE	This array contains all of the wavelengths that are to be used with each window. If the wavelength is given then the window is processed ONCE and then the program goes to the next window. If the wavelength is zero then the window is processed with a wavelength that is calculated. Also, if the program detects that there is more noise it will process the window again and keep doing this until either the program detects no more noise or four passes have been made through the window.
FILT	This parameter is used when the wavelength parameter is zero. This is the best guess wavelength parameter. It is used to help find the wavelength of the noise. Note that this parameter is the UPPER LIMIT of the wavelengths, no calculated wavelength can be greater than this value.