The last required file is an IBIS file giving information on the input
files. The columns of the file are usually:
1. The name of the image file in any order, with or without pathname.
Only the part of the pathname/filename after the last \ or / is read
for checking
2. The offset SL of the image input
3. The offset SS of the image input
4. The NL of the image input (must be exact, cannot be less as in FASTMOS)
5. The NS of the image input (must be exact, cannot be less as in FASTMOS)
6. Input factor for brightness equalization formulas (only one as of 9/99)
7. Reserved for more complex brightness equalization formulas
the numbering of the columns can be changed by the parameter COLS. See
parameter ADJUST for an explanation of column six.
The program limits at present (9/99):
1. 400 inputs
2. The number allowed by the system for simultaneous open (somewhere
between 67 and 100 according to one expert).
The files are opened and closed as a narrow mosaicking band passes
from top to bottom of the output. The width of this band is DFEATHER.
The number of files open over the band cannot exceed the VICAR
or system limit for number of files open at once. It is possible that
a UNIX system call can enlarge this.
The optional GeoTIFF reference image, which is placed after the IBIS file,
is used in the following way:
1. Its pixel referencing is a master for the mosaic.
2. All other images will have offsets relative to the master.
3. The SIZE or SL,SS,NL,NS parameters will be relative to the master.
4. The offsets can be placed in the IBIS file, in which case they are
checked against the GeoTIFF mappings.
5. The offsets can be zero'd in the IBIS file, in which case they are
calculated from the GeoTIFF mappings. NL,NS columns must be zero'd
for this to occur and they are also calculated.
6. For this case, the 'GEOTIFF keyword must be given and all input images
are required to have GeoTIFF labels.
A major difference with program FASTMOS is that (SL,SS) can be used to window down into the mosaic. See HELP level 2 under SL.
The last file is an IBIS file giving information on the input files. The
columns of the file are usually:
1. The name of the image file in any order, with or without pathname.
Only the part of the pathname/filename after the last \ or / is read
for checking
2. The offset SL of the image input
3. The offset SS of the image input
4. The NL of the image input (must be exact, cannot be less as in FASTMOS)
5. The NS of the image input (must be exact, cannot be less as in FASTMOS)
6. Input factor for brightness equalization formulas (only one as of 9/99)
7. Reserved for more complex brightness equalization formulas
By using this parameter, the columns can be renumbered as the user desires.
If the SIZE field is not entered, the output image has the same size as the first input image. If the SIZE field is entered, the number of lines and number of samples refer to the size of the output image.
The default is 1. Setting it larger moves the "window" of the mosaic upwards in line number and is the same as lowering all of the individual input SL's by an equivalent amount. This means that the mosaic can conveniently be done in sections by setting (SL,SS,NL,NS) in a checkerboard fashion and then by butting together the checkerboard pieces with the VICAR programs APPEND and MSS.
The default is 1. Setting it larger moves the "window" of the mosaic upwards in sample number and is the same as lowering all of the individual input SS's by an equivalent amount. This means that the mosaic can conveniently be done in sections by setting (SL,SS,NL,NS) in a checkerboard fashion and then by butting together the checkerboard pieces with the VICAR programs APPEND and MSS.
The mosaicking mode specifies how the output data number values are determined
from the input data numbers. The following rules apply for each of the
modes. If none of the input images have a data number value for a
given pixel that is greater than or equal to the THRESH value, the
output data number is the data number from the first input image if
the pixel is contained in the first input image, and the output
data number is L0 if the pixel is not contained in the first input image, where
L0 is 0 if THRESH is greater than 0, L0 is 0 if THRESH=0 and the data format
is byte, and L0 is equal to THRESH-1 otherwise. If
exactly one of the input images has a data number value for a given pixel that
is greater than or equal to the THRESH value, the output data number is the
data number from the one input image. If more than one of the input images
have a data number value for a given pixel that is greater than or equal to the
THRESH value, the output data number is determined by the mosaicking mode.
There are currently five modes to choose from. They are listed by name below.
For each mode a description is given of how the output data number is
determined when there is more than one input image having a data number value
that is greater than or equal to the THRESH value for a given pixel. The
default mode is OVERLAY.
OVERLAY - The input images are checked against the THRESH value in the order
in which they are entered by the user. The first data number value
found which meets the threshold criteria is used for the output
image. This means that the order in which the input files are
entered gives a priority to the data in the files.
AVERAGE - The average of the values meeting the threshold criteria is used.
The average is found by integer division with no rounding.
MOD - When there are two values meeting the threshold criteria, the
average of the values is used. When there are more than two
values meeting the threshold criteria, the value closest to the
average is used. This mode may be particularly useful when
combining many images with high bit-error rates.
MAX - The maximum of the values meeting the threshold criteria is used.
MIN - The minimum of the values meeting the threshold criteria is used.
Only values greater than or equal to the THRESH threshold parameter are used by the mosaicking mode in determining the data numbers for the output image. The THRESH value is usually greater than 0 for mosaicking. THRESH can be set to 0 for cases such as averaging images. The default value is 1. (See Example 5 in the main help for details about the case of THRESH=0 for byte data. Users may need to convert images to halfword to use THRESH=0 for mosaicking. Other VICAR programs, such as INSECT may be an alternative.) For halfword images for which negative DNs are considered valid, a negative THRESH value may be used. In this case, 0 is an inappropriate value for representing the absence of image data. When THRESH is less than 0, FEATHERV uses an output DN of THRESH-1 to represent the absence of image data. (If THRESH = -32768, -32768 is used to represent the absence of image data.) This value is used as an output DN where the output pixel does not lie in one of the input images. (See the MMODE parameter.)
If the PROGRESS parameter is specified, FEATHERV prints a comment every 500th line. The default is to not print the progress.
If the EDGE parameter is specified, all input images are edged prior to applying the mosaicking mode. No edging is the default. 'EDGE M U S T be specified to invoke the edging algorithm. ALSO NOTE THAT EDGING IS INCREASED BY THE USE OF MOOREFAC. A NIBBLE OF TWO AND A MOOREFAC OF THREE WILL RESULT IN SIX PIXELS BEING REMOVED. If edging is selected, then each line of each input image is edged first on the left and then on the right. Edging means scanning through a line from one end or another to the point at which the data numbers are greater than or equal to a threshold value and then removing the pixels from the end of the line up to a certain number of pixels beyond the point. This is referred to as finding the edge of the scene data and nibbling in a certain number of pixels beyond the edge. Because of the line by line processing performed by the program, edging is only available in the horizontal direction. (Edging the top and bottom of images would normally require an intermediate data set.) Edging is typically used to remove distortion around the edges of pictures that was caused by interpolation, filtering, or other things. Several parameters are used to control the way that edging is done. The parameters NTHRESH, LTHRESH, RTHRESH, NSEQ, and NINCR determine the location of the edge of the scene data for lines of the input images. The parameters NIBBLE, LNIBBLE, and RNIBBLE determine how many pixels beyond the edge of the scene data are removed. The edge of the scene data for a line is determined as follows. The program scans through the pixels of a line comparing the data numbers against the edging threshold. (Separate edging thresholds can be specified for scanning from the left and scanning from the right using the LTHRESH and RTHRESH parameters. The NTHRESH parameter can be used to specify the same threshold for scanning from the left and scanning from the right.) The scanning begins at one end of the line, and it checks successive pixels unless the NINCR parameter is entered. If NINCR is entered, the scanning checks only every NINCRth pixel. The program scans until it finds a group of NSEQ consecutive (in terms of NINCR) pixels all of which have a data number greater than or equal to the edging threshold. The edge of the scene data is defined as the first pixel (according to the direction of the scan) of that group. The nibbling number is the number of pixels, starting with the edge of the scene data, which are to be removed along with any pixels from the end of the line to the edge of the scene data. (If the nibbling number is zero, then just the pixels from the end of the line to the edge of the scene data are removed.) Separate nibbling numbers can be specified for scanning from the left and scanning from the right using the LNIBBLE and RNIBBLE parameters. The NIBBLE parameter can be used to specify the same nibbling number for scanning from the left and scanning from the right. If no edge of the scene data is found when scanning, the entire line is removed.
The default for NTHRESH is THRESH. (See also under EDGE.)
The default for LTHRESH is NTHRESH. (See also under EDGE.)
The default for RTHRESH is NTHRESH. (See also under EDGE.)
The default for NSEQ is 8. (See also under EDGE.)
The default for NIBBLE is 4. (See also under EDGE.) ALSO NOTE THAT EDGING IS INCREASED BY THE USE OF MOOREFAC. A NIBBLE OF TWO AND A MOOREFAC OF THREE WILL RESULT IN SIX PIXELS BEING REMOVED.
The default for LNIBBLE is 4. (See also under EDGE.) ALSO NOTE THAT EDGING IS INCREASED BY THE USE OF MOOREFAC. A NIBBLE OF TWO AND A MOOREFAC OF THREE WILL RESULT IN SIX PIXELS BEING REMOVED.
The default for RNIBBLE is 4. (See also under EDGE.) ALSO NOTE THAT EDGING IS INCREASED BY THE USE OF MOOREFAC. A NIBBLE OF TWO AND A MOOREFAC OF THREE WILL RESULT IN SIX PIXELS BEING REMOVED.
The default for NINCR is 1. (See also under EDGE.)
This keyword is defaulted to 'NOADJ which does no brightness adjustment. Three kinds of adjustment can be performed using column 6 of the IBIS file: 'FACTOR - All input images will be multiplied by the number in column 6 of the IBIS file as a brightness adjustment. Note that there is a mathematical risk of a one turing into a zero if the factor is less than .5. This would be significant if zero means "no data" for mosaicking. 'ADD - All input images will be added to by the number in column 6 of the IBIS file as a brightness adjustment. 'ADDZMASK - All input images will be added to by the number in column 6 of the IBIS file as a brightness adjustment except that zeros/nonzeros are preserved (pixels that were greater than 0 prior to the add are compared to a minimum of 1, zeros remain at zero). This is useful for the case where zero means "no data" for mosaicking. The adjustment is performed just prior to the output or the averaging for output, so it does not affect the edging operations. The cloud erasing operation is done after this adjustment operation.
Width of feathering is the max distance of the MOORE DISTANCE ALGORITHM measured in pixels. The effective feathering will be twice this distance because a large overlap will feather one image to its full MOORE DISTANCE yielding a 50/50 averaging of the two images. Then a reverse feather of the other image will take place. A larger number will feather the inputs to a greater degree, but the program will use more time and memory. Plan carefully using the virtual memory limit formula 6 x (max number of inputs open) x (max input NS) x DFEATHER / MOOREFAC**2 If MOOREFAC is greater than one, the value used is DFEATHER/MOOREFAC in tiles of size MOOREFAC x MOOREFAC, and also the DFEATHER will be adjusted slightly to a multiple of MOOREFAC because of the tiling. The net effect of MOOREFAC>1 is to still feather to depth DFEATHER, but in coarser steps.
In the formula for storage: 6 x (max number of inputs open) x (max input NS) x DFEATHER / MOOREFAC**2 DFEATHER increases storage linearly, and also increases compute time. The MOOREFAC reduces the resolution of the MOORE algorithm and decreases storage quadratically, and also saves compute time. Further, the ratio of DFEATHER to MOOREFAC must be less than 65536. A good suggested value is a ratio of 50 to 100. This will result in 100 to 200 steps of brightness in feathering from one image to the other. If ramping is used, the need for feathering is reduced and a smaller ratio can be used. Keep in mind that increasing MOOREFAC does not decrease the feathering distance in the images, it simply coarsens the number of steps in the MOORE distance function. The human eye probably cannot see more than 100 steps in the typical mosaicking situation.
If a pixel comes from multiple inputs and one of the inputs is brighter than one of the others (which of the others is order dependent for three or more) by this amount, then it is deleted from the averaging process. This correction is applied after the brightness adjustment (see keyword ADJUST).
'RAMP - apply ramping procedure; The first pass over the files calculates the ramp values to apply to each scene. The second pass constructs ramps to smooth together the scenes. The ramp values are written to the IBIS file. 'NORAMP - don't apply ramping A single pass smoothes together the scenes using the Moore distance interpolation only. No ramp values are calculated. No ramp values are written to the IBIS file. 'READRAMP - apply previous ramp values from file This option requires an IBIS file with ramp values calculated from a previous run with the 'RAMP option. The previous run can be on a reduced image... the ramp values are size-independent. A single pass is performed like the second pass of the 'RAMP option, and no values are written to the IBIS file.
This is for throwing out unreliable water or shadow pixels in Landsat, or space, shadow, etc. for planets.
This allows glimmer, seasonal variation, wetness, etc. to be discarded from tiepoints. It is applied after the adjust parameter correction.
Tiepoints in overlap areas for an image are taken at a particular distance inside of the neighbor image. The Moore distance function is used to determine when a pixel is that distance inside. The unit of distance is the pixel. By setting this parameter to 7 (for example) all pixels that are a distance of 7 from the edge of non-0 data in the neighbor overlap will be used as tiepoints. The tiepoints from different neighbor frames are kept separate in the tally, so that brightness ramps can be constructed to each neighbor. Neighbors are specified in the IBIS file input (see parameter rcols).
The ramps are four half-images hinged horizontally and vertically in the middle of an image. The difference with a neighbor (divided by 2) is the height of the ramp at the edge of the image. But the data may actually start inside the image with a zero gore of data at the edge. This factor multiplies the height of the half plane so that it applies better to the data.
The ten columns in order are (using their default column numbers, which can be changed)
8. An index of the images (any unique integer index will do)
9. Right neighbor to ramp to, or 0 if no neighbor to the right (using the index
of column 8)
10. Bottom neighbor to ramp to, or 0 if no neighbor to the bottom (using the index
of column 8)
11. Left neighbor to ramp to, or 0 if no neighbor to the left (using the index
of column 8)
12. Top neighbor to ramp to, or 0 if no neighbor to the top (using the index
of column 8)
13. Constant calculated by the ramping procedure, add to this image
14. Ramp constant calculated by the ramping procedure for right-facing ramp (a
value of -999.0 means that there is no ramp in this direction##)
15. Ramp constant calculated by the ramping procedure for bottom-facing ramp (a
value of -999.0 means that there is no ramp in this direction##)
16. Ramp constant calculated by the ramping procedure for left-facing ramp (a
value of -999.0 means that there is no ramp in this direction##)
17. Ramp constant calculated by the ramping procedure for top-facing ramp (a
value of -999.0 means that there is no ramp in this direction##)
## A ramp of height zero is different than no ramp. The ramp of height zero
would be interpolated with neighboring ramps, causing them to curve towards
zero. No ramp (-999.0) would not be interpolated with neighboring ramps.
GeoTIFF Options: If the images all have GeoTIFF labels and the 'GEOTIFF keyword is given, then there are two possibilities: The first is: if columns 2-5 contain non-zero values (only column 4 is checked for 0), then the values are compared with values calculated from the GeoTIFF labels or the nl,ns from the VICAR label. Since the values might calculate with a slight inaccuracy, the TOLER parameter is provided with a default value of 1.e-7. If a calculated offset exceeds the TOLER parameter, this indicates that the files have an inaccurate mapping for the purposes of mosaicking and do not have an integral offset. All mappings are checked for consistency (map projection type, zones, meridians, etc.). The second is: if columns 2-5 contain zero values (only column 4 is checked for 0), then the values are calculated from the GeoTIFF labels or the nl,ns from the VICAR label. The only check is that the offsets are integral values (within the parameter TOLER) and that the mappings are consistent.
The calculated value, if within TOLER of an integral value, will be converted to the integral value for purposes of mosaicking.
FEATHERV HAS BOTH THE OLD TYPES OF NIBBLING (SEE PARAM EDGE) AND A NEW EDGING CAPABILITY BASED ON THE MOORE ALGORITHM. The new edging nibbles in a perpendicular direction to the edge of data in all directions (the old nibble worked from the righ or left only). Use the parameter MOORENBL to invoke this. The default of 0 is no nibble, a value of n nibbles all values where the MOORE distance from the edge is n, etc. To give a smooth transition, instead of starting at n+1, the whole MOORE distance input is reduced by n so the MOORE distance that is used in feathering still starts at 1. Keep in mind that MOOREFAC=3 means that a MOORENBL=1 will nibble 3 pixels from all edges.
This will help if the images do not register perfectly in the geometric (x,y) sense. A road that appears doubled in the overlap area will turn into a disjoint line. The seam will be along the polyline where the Moore numbers are at equality. Do not use this parameter along with MOOREPOW. FEATHERV has two parameters to adjust the mathematics of the feathering calculation. MOOREMAX and MOOREPOW. The choice here is affected by the quality of image geometric registration. If the images are not in perfect registration, consider using MOOREMAX. If MOOREMAX causes an undesirable brightness edge, consider using MOOREPOW with a high number, lowering it until the brightness edge disappears.
If set to a number greater thatn the default 1.0, the Moore distance values will be taken to this power. So the linear ratio of Moore numbers, say, 80 to 20 would be converted into 80**pow to 20**pow (or a pow split FEATHERV has two parameters to adjust the mathematics of the feathering calculation. MOOREMAX and MOOREPOW. The choice here is affected by the quality of image geometric registration. If the images are not in perfect registration, consider using MOOREMAX. If MOOREMAX causes an undesirable brightness edge, consider using MOOREPOW with a high number, lowering it until the brightness edge disappears.