The horizontal accuracy of the orthrectified imagery is primarily determined by the quality
of the Aerial
Triangulation solution and the surface model (DEM).
All ground targets were visible with .2 meter x 3.6 meter targets, and measured with GPS.
The AT procedure takes the raw bands (BW,FW and N) and rectifies them to a plane using the
GPS data that was collected simultaneously with the imagery. This was accomplished using the
Leica Gpro software. Automatic Point Measurement (APM) was used to place terrain points on
each panchromatic L1 image. The panchromatic L1's along with the each of the colour L0's were
then imported into ORIMA so that an AT adjustment could be executed. At this stage each
individual strip of imagery was not tied to the neighbouring strips. Several iterations were
run until all bad points were eliminated. The points were then exported to an x,y,z text file
so that a DEM could be created, allowing APM a surface file so that it could more accurately
calculate the tie points. Once the second APM process was complete, the strips were tied
together, and several adjustments were run, removing the points with high residuals until a
suitable sigma value was reached. Control was then measured and a final adjustment
calculated. Standard deviations and residual were reviewed to verify that it was a good
solution. The support files were then updated with the new orientation information. The L1
RGB bands were then rectified using the updated orientation information.
Specifically for SLC, cross-strips running E-W were used in the AT solution to give strength
to the block being worked. Between blocks, there was a common cross strip used in the AT
giving overall strength to the solution.
The Data Capture stage was accomplished by ingesting the Orima AT solution into the ISPM Z/I
software. Using an Z/I SSK workstation and ISSD software for viewing, The surface data was
collected with in-house VMS data cature software supplimenting a USGS surface with DEM points
and breaklines.
The rectification stage involved rectifying the 16 bit per channel imagery using Z/I's ISBR
software. We then rectified the L1 data into manageable pieces with overlapping imagery
between the strips, converted the imagery to 8 bits per channel. The radiometry was then
adjusted, the rectified imagery was run through our in-house AIM software for automated
seamline generation. These seamlines were then checked, and once manually verified, the final
ortho tiles were generated. At the QC stage any building or
above ground feature obscuring a transportation feature is corrected. All orthophoto tiles
were checked visually for relative horizontal accuracy in reference to adjacent tiles.
Orthophoto measurements were taken from the 6 horizontal ground control points in the project
area. An RMSE was calculated for the XY error residuals of the positional differences between
the surveyed values and the values measured from the rectified imagery.
Once received by the UGRC the images were mosaiced into groups that on average are 9 of
the original GeoTIFF images.The process used to mosaic the images was to turn the images
in multi-stack grids, mosaic the grids for each individual band in GRID, make a stack of
the merged bands, and turn the new grid stack back into an image.