This section provides descriptions of the modes of processing data geometrically. This includes the process of rectifying an image to a map base, registering an image to another image, and changing projections of an image. These examples are provided to guide the firsttime or occasional user through the registration process. Many other combinations of modules are possible; the most direct method for each type of registration is given.
The process of rectifying an uncorrected image (the search image) to a map base (the reference space) proceeds as follows.
LAS 7.0 contains various modules for ingesting image data. Many of these modules place nominal location and projection information in the image's DDR. This information is very useful during the registration process, even if it is not exact. Examples of these modules are ccttipsp for TM-P data, dementer for DEM (elevation) data, lacin for AVHRR data, and edipsin and plenter for MSS data.
Image preparation refers to the fixing of line drops (fixlin) or other image imperfections. It may also include contrast stretches (map) to make the image data more manageable for the tie point selection process.
The interactive selection of ground control point data from a map is accomplished using the tiepts module. Output consists of two tie point selection files, one containing geographic coordinates from the map(s), typically named ref;tps, and one containing the search image coordinates, typically named srch;tps, of the corresponding points. Note: Appendix B describes the digitizer interface and Appendix C describes subpixel image coordinates.
%tiepts
The geographic coordinates in ref;tps must be converted to line and sample locations in the output image space. To accomplish this, three modules are used. First, the output image projection parameters are established using the projprm module. In this example, the output image will be registered to a Transverse Mercator projection with a Clark 1866 ellipsoid, a scale factor at the central meridian of 0.9996, a central meridian of 103 degrees west, and with a latitude of origin of zero. The unit of measure of the projection is meters.
LAS> PROJPRM-TM OUTPROJ=EXAMPLE PROJKEY="TM" +
SCALFACT=.9996 CENTMER=-103.0 ORIGIN=0.0 +
GEOUNITS=DEG DATUM=0 FALEAST=0 FALNORTH=0 +
PROJUNIT=MET
Next, the output image space frame must be established and the geographic coordinates in ref;tps gridded to output image space coordinates. This is accomplished using the trancoord module and the -grid subcommand. The output space is framed using the Transverse Mercator projection previously described, which includes the area bounded by 41.35 north, 107.75 west and 37.90 north and 100.10 west. Each pixel is 50 meters square (in the projection distance). Refer to Appendix D for more information on the framing process.
LAS> TRANCOORD-GRID IN=-- INTS=REF +
INPROJ=EXAMPLE IPROJKEY=-- OPROJKEY="TM" +
OUTTS=REF.MAPPED PIXSIZ=(50,50) +
COORS=(41.35,-107.75,37.90,-100.10) +
COORUNIT="DEG" PRINT=TERM
Finally, the two tie point selection files, ref.mapped;tps and srch;tps, are merged into a tie point location file, example;tpl, for use in the modeling process.
LAS> TIEMERGE-NOCORR INTS=(REF.MAPPED,SRCH) +
OUTTL=EXAMPLE CONFLG=NO
Example;tpl contains line and sample coordinates for both the search image and the resulting output image space. These coordinate pairs are now used to derive a pair of second order least squares regression equations that predict input coordinates from output coordinates. Results of the predictions are displayed, including point identifiers, coordinates, residuals of the transformation, and RMS residual values.
The user may remove or reinstate points and refit until a satisfactory fit is obtained. When a satisfactory fit has been obtained, a geometric mapping grid is created.
LAS> POLYFIT INTL=EXAMPLE WINDOW=-- +OUTTL=EXAMPLE.EDIT OUTGRID=GRID PDEGREE=2
The input image, srch;img, is now rectified to the map base defined by the geometric mapping grid grid;grid. Cubic convolution with an alpha parameter of -0.5 is the resampling method used. The output image size was determined during the process of framing the output image space and is contained in the geometric mapping grid.
LAS> RESAMPLE-GRID IN=SRCH INGRID=GRID OUT=IMAGE.OUT +
RESAMP=CC
An alternate way to perform this rectification combines the steps of coordinate transformation and framing, point modeling and grid generation, and geometric transformation steps into one module-- register. This method is less flexible than running the functions individually but is useful for the occasional user. This is accomplished by running tiepts, projprm, and register.
%tiepts
LAS> PROJPRM-TM OUTPROJ=EXAMPLE PROJKEY="TM" +
SCALFACT=.9996 CENTMER=-103.0 ORIGIN=0.0 +
GEOUNITS=DEG DATUM=0 FALEAST=0 FALNORTH=0 +
PROJUNIT=MET
LAS> REGISTER-IMG2MAP IN=SRCH INTS=(REF,SRCH) +
INPROJ=EXAMPLE OUT=IMAGE.OUT IPROJKEY=-- +
OPROJKEY=TM PIXSIZ=(50.0,50.0) +
COORS=(41.35,-107.75,37.90,-100.10) +
COORUNIT="DEG" PDEGREE=2 RESAMP=CC +
CONFLG=NO PRINT=LP
The process of registering an image (the search image) to another image (the reference image) proceeds as follows.
LAS 7.0 contains various modules for ingesting image data. Many of these modules place nominal location and projection information in the image's DDR. This information is very useful during the registration process, even if it is not exact. Examples of these modules are ccttipsp for TMP data, dementer for DEM (elevation) data, lacin for AVHRR data, and edipsin and plenter for MSS data.
Image preparation refers to the fixing of line drops (fixlin) or other image imperfections. It may also include contrast stretches (map) to make the image data more manageable for the tie point selection process.
The interactive selection of tie point data is accomplished using the tiepts module. Refer to the tiepts user guide for a description of the point selection process. Points may be selected manually or automatically after a number of manual points have been selected. Output consists of two tie point selection files, one containing reference image coordinates and the other containing search image coordinates.
% tiepts
Next, the two tie point selection files (typically named ref;tps and srch;tps) are merged into either a merged tie point file for use in the correlation process or a tie point location file for use in the modeling process.
If image-to-image correlation is not used, create the tie point location file example;tpl.
LAS> TIEMERGE-NOCORR INTS=(REF,SRCH) OUTTL=EXAMPLE
+ CONFLG=NO
Otherwise, create the merged tie point file example;mtp for use in the correlation process.
LAS> TIEMERGE-PRECORR INTS=(REF,SRCH) + OUTMT=EXAMPLE
CONFLG=NO
The correlation process refines tie points using either grey level or edge correlation. This step may be skipped if the user thinks tie points were selected with satisfactory accuracy. If tie points were collected in automatic mode with tiepts, correlation is recommended. Output from the correlation process is a tie point location file, example;tpl.
LAS> CORRELATE-GREY IN=(REF,SRCH) INMT=EXAMPLE +
OUTTL=EXAMPLE MINCORR=6.0 MAXDIFF=-2.0 +
FITMETH=PARAB CONFLG=NO
Example;tpl contains line and sample coordinates for both the search and reference images. These coordinate pairs are now used to derive a pair of second order least squares regression equations that predict input coordinates from output coordinates. Results of the predictions are displayed, including point identifiers, coordinates, residuals of the transformation, and RMS residual values.
The user may remove or reinstate points and refit until a satisfactory fit is obtained. When a satisfactory fit has been obtained, a geometric mapping grid is created.
LAS> POLYFIT INTL=EXAMPLE WINDOW=(1,1,2000,2000) +
OUTTL=EXAMPLE.EDIT OUTGRID=GRID PDEGREE=2
The input image, srch;img, is now registered to the reference image space defined by the geometric mapping grid grid;grid. Cubic convolution with an alpha parameter of -0.5 is the resampling method used. The output image size was determined during the modeling process (polyfit module/window parameter) and is contained in the geometric mapping grid.
LAS> RESAMPLE-GRID IN=SRCH INGRID=GRID OUT=IMAGE.OUT +
RESAMP=CC
An alternate way to perform this registration combines the steps of tie point merging, correlation, point
modeling and grid generation, and geometric transformation steps into one module, register. This
method is less flexible than running the functions individually but is useful for the occasional user. The
user thus runs tiepts and register.
% tiepts
LAS> REGISTER-IMG2IMG IN=SRCH INTS=(REF,SRCH) +
INPROJ=EXAMPLE OUT=IMAGE.OUT +
WINDOW=(1,1,2000,2000) CORRMODE=GREY +
BANDS=(1,1) PDEGREE=2 RESAMP=CC CONFLG=NO +
PRINT=LP
2.3 Changing Image Projections
The projection of an image is changed using the projprm, remap, and resample functions. Remap
operates in two modes--one mode reads input (existing) projection information from the input image
DDR and the other obtains it manually from parameter input. In this example, a DEM image with valid
projection and location information in its DDR is being reprojected to match the image rectified in the
image-to-map example above. Projprm defines the output projection system, remap determines image
frame and creates a geometric mapping grid, and resample remaps (transforms) the input image using the
geometric mapping grid and bilinear interpolation resampling.
LAS> PROJPRM-TM OUTPROJ=EXAMPLE PROJKEY="TM" +
SCALFACT=.9996 CENTMER=-103.0 ORIGIN=0.0 +
GEOUNITS=DEG DATUM=0 FALEAST=0 FALNORTH=0 +
PROJUNIT=MET
LAS> REMAP IN=DEM INPROJ=EXAMPLE OUTGRID=GRID +
OPROJKEY="TM" PIXSIZ=(50,50) +
COORS=(41.35,-107.75,37.90,-100.10) +
COORUNIT="DEG" ROTANG=0.0 PRINT=TERM
LAS> RESAMPLE-GRID IN=DEM INGRID=GRID OUT=DEM.OUT +
RESAMP=BI
Another scenario involving the changing of image projection is the combination (side-by-side) of two
data sets which are registered to a map projection but not to the same map projection. As an example,
consider two DEM images with valid DDRs, one registered to UTM zone 11 and one registered to UTM
zone 12. Before these two images can be combined, they must be in a common projection system. In
this example, the DEM registered to UTM zone 12 will be remapped to UTM zone 11. The two images
may then be combined using the concat function:
LAS> PROJPRM-UTM OUTPROJ=EXAMPLE +
PROJKEY="UTM11" ZONENUM=11
LAS> REMAP IN=DEM.12 INPROJ=EXAMPLE OUTGRID=GRID#+
OPROJKEY="UTM11" PIXSIZ=(30,30)
LAS> RESAMPLE-GRID IN=DEM.12 INGRID=GRID OUT=DEM.12.11 +
RESAMP=BI
LAS> CONCAT-AUTO IN=(DEM.11,DEM.12.11) OUT=DEM.MERGE