Next: Inputfile
Up: Introduction
Previous: Processing order
Contents
In this section some important definitions are described that are used
in the Doris software. This will clarify the terminlogy used. The
general set up of the input file and output files is described as
well.
After compilation with the Makefile, the executable is named:
"doris". In this document therefor this name is used to refer to the
executable. The command line options are:
- doris -ver
return version number.
- doris -h
return help (system call to shell script named helpdoris).
- doris [file]
run, use input in ''file'' (default: ''inputoptionsfile''.
It is advised also to compile an executable ''doris.debug'' (with the
Makefile). This version is somewhat slower and more verbose, but it
can be used if something seems to go wrong with the normal executable,
and it is not clear what. See also Annex B.
We advice to use the utility scripts to generate input files and to
run the processor.
Conventions are:
- We use the term lines to refer to the azimuth direction
(slow time), and pixels for the range direction (fast time).
(A pixel might also refer to an element, which will be clear
from the context.) In the source code we frequently use the term
azimuth buffers and range blocks.
- Our convention is to use first lines, second pixels,
e.g., for the order of input arguments. The line direction
(azimuth) corresponds to the vertical (y). The pixel direction
(range) corresponds to the horizontal (x). Note that other software
may use x before y.
- The first line (pixel) of an image is indexed as 1 (this may be
a bit unusual). (In the software, the first index of an image (in a
matrix) is equal to 0. To index a matrix, use MAT(y,x), i.e., as in
linear algebra, Matlab, etc.)
- The name, format, and dimensions of the current
master/slave/interferogram are stored in information structs that
are filled by reading the corresponding result files.
The files do not have a header.
- Generally all coordinates are in the master (radar) coordinate
system. The first and last line are given, as well as multilook
factors for both directions. If for example an interferogram is
multilooked at the generation with a certain factor and later on
again at the subtraction of the reference phase, then the first line
is still the same, see Figure 1.3.
- In our view, the output files can only contain the results of
one algorithm per step. So it is not possible to re-run a step (for
example with another algorithm) without deleting the previous
result. New users may be confused by this approach, but Doris
should give an appropriate error message if it is attempted to run a
step twice.
- Temporary files are created during the processing. Their names
always start with "scratch". If such files are not removed by
Doris, for example after an error in the processing, they can be
savely removed by a rm command.
Figure 1.3:
Principle of multilooking.
|
In the logfile additional information is written that is not sent to
standard out or the result files, such as statistical information on a
least squares estimate. (For example, we always inspect the
correlation value for step COARSE_CORR, and the error between model
and observations for reference phase computations in the log file.)
Subsections
Next: Inputfile
Up: Introduction
Previous: Processing order
Contents
Leijen
2009-04-14