Help for DS4

PURPOSE:
DS4 attempts to remove 6 line striping from LANDSAT imagery by
performing localized histogram matching of the 6 sensors.

EXECUTION:

Examples
	DS4 IN OUT

	will remove striping using sensor 3 as the reference sensor
	(which is the default), and assuming 3 sets per group (default).
	The first line of the first set, in this case, is line number 1,
	chosen so that the reference sensor becomes the third line within
	the set.

	DS4 IN OUT RSEN=5

	will remove striping as above, with the exception that the
	reference sensor will be sensor number 5, making line number 3
	the first line of the first set.

	DS4 IN OUT 'AVERAGE

	will remove striping as in the first example above, but instead
	of using one particular sensor as a reference sensor, will
	use the average of all of the sensors as a reference.  Line 1
	is again the first line in the first set.

	DS4 IN OUT 'AVERAGE RSEN=6

	Here the AVERAGE option has been combined with the RSEN option
	to cause averaging and, at the same time, make sensor number 6
	the third line in the image.  The net effect is to ignore the
	first three lines of the first set.  Again, as above, the
	number of lines per group defaults to 3.

	DS4 IN OUT FILTER=(1,2,3,2,1) RSEN=1

	The reference sensor is sensor number 1, and there are three sets
	in a group.  Line 5, then, is the first line of the first set.
	The reference CDF is filtered with the weights 1,2,3,2,1 before
	matching is performed.

	DS4 IN OUT RSEN=2 GROUP=7 'AVERAGE  FILTER=(1,2,1)

	Averaging and filtering are performed as above, and line number 6
	is the first line of the first set.  Note, here, that there are
	7 sets per group.

SPECIAL NOTES:
1) LANDSAT frames have been received with 12 line striping superimposed
   upon the 6 line striping.  DS4 will not correct the 12 line striping.
   The 12 line striping should be removed by other means prior to the
   execution of DS4.
2) DS4 is unable to remove 6 line striping from water bodies.


OPERATION:
The input picture is considered to be subdivided into contiguous sets of
6 lines each.  The starting line of a given set is determined by the
reference sensor number such that the reference sensor becomes the third
line within the set.  For example, if the reference sensor were chosen to
be 5, then the starting line of the first set would be 3 and that of the
second set would be 9, and so on.  Note that the line numbers above
refer to the image given by the VICAR size field.  That is, if a subarea
is specified, then line 5 above refers to the fifth line of the subarea.

To initiate the matching process, the CDF's for each sensor are compiled
using the first group of sets (i.e., the first GROUP sets).  The reference
CDF is that of the specified sensor RSEN.  If the AVERAGE parameter is
specified, the reference CDF is taken to be the average of all 6 sensor
CDF's.  If the parameter FILTER is specified, then the reference CDF
is low-pass filtered using the integer weights F1, . . ., FN.  The CDF
of each sensor is matched to that of the reference CDF and corresponding
transfer tables are generated.  The first ((N2-1)/2) + 1 sets plus any
lines skipped over at the beginning (see the first paragraph) are then
transformed using the generated transfer table.

Subsequent sets are transformed in the following manner:
1) The CDF's of the sensors of the first set in the group are subtracted
   out of the group sensor CDF's.
2) The CDF's of the sensors in the first set following the group are added
   in to the group sensor CDF's.
3) The transfer tables are generated as before.
4) The middle set of the updated group is transformed.

This process continues until the last set of the image has been included
in a group.  At this point, the middle set of the group, all sets following
it, and any lines after the last set are transformed.

The parameters AVERAGE and RSEN are not mutually exclusive.  While the
AVERAGE parameter will override the RSEN parameter in terms of how the
reference CDF is generated (i.e., from one sensor or from an average of all
six sensors), the RSEN parameter can be used to define which lines
compose the first set of lines and hence all subsequent sets of lines.
This was illustrated in the examples.

PROGRAM HISTORY

WRITTEN BY:  Daryl Madura, 5 January 1979
COGNIZANT PROGRAMMER:  A. S. Mazer
REVISIONS
SEP 02 83  Alan Mazer  MODIFIED FOR VAX CONVERSION
JUN XX 87  Ron Alley   CONVERTED TO VICAR2 BY RON ALLEY
MAR 15 98  GMY  Ported to Unix