4.0 Description of a software system for Terrain
Mapping Using IRS-1C Imagery
The software system developed at Space
Applications Centre (ISRO) for terrain mapping
using IRS-1C imagery was used for analysis of
IRS-1C capabilities. This system consists of the
following basic elements:
(i) Ground Control Point (GCP) tools
(ii) Digital Elevation Model (DEM) generation
and editing
(iii) Orthoimage generation
(iv) Quality Evaluation and
(v) Mapping using GIS
All these are additionally supported by image
processing and graphics libraries. The hardware
consists of an Indigo-2 R-4000 based softcopy
photogrammetry workstation with stereo display
monitor, crystal eye glasses for stereo view, a
scanner and a plotter. The details of the software
system are explained in the subsequent sections.
4.1 GCP Generation
Basically this consists of a scanner system, wherein
the desired maps are scanned in an appropriate
resolution. The digital image is properly
thresholded, so that all the linear features are
clearly identifiable. The scanned image is then
transferred to the main system for GCP
identification along with the stereo pair images.
Co-ordinates of a point in terms of latitude and
longitude is obtained by a projective transformation
on the surrounding known grid points of the map.
The height of the point is interpolated from the
nearest elevation contour lines. The image co-
ordinates of the control points are identified
parallelly on the display device.
4.2 DEM generation and editing
A DEM is regarded as a numerical description of the
surface of an object on measured or derived co-
ordinates of numerous scattered points. With the
advent of digital phtogrammetry workstations with
high computation power, DEMs can be derived by
purely digital approach, from the satellite stereo
pairs. The digital mode of DEM generation from a
satellite stereo pair consists of the following steps
(i) Automatic identification of conjugate points
(ii) Determination of satellite orientation using
a model based on orbit and attitude parameters
(iii) Three dimensional co-ordinate
determination by the method of space intersection
(iv) Bundle Adjustment
(v) Height interpolation to compute heights at
regularly spaced grid points and
(vi) Point editing to remove the spurious height
points and reinterpolation
4.3 Orthoimage Generation
Basically a precise ground to image relation along
with the DEM and the raw data are the requirements
to generate an Orthoimage. This is a geocoded
product corrected for all the geometric errors
including terrain relief and the camera tilt, which can
be directly used for topographic mapping. The
Orthoimage can directly go into a. GIS. For an
output grid of latitude and longitude and height
(obtained from DEM) time and pixel of the input
image can be calculated by an iterative way using
equations (2),(3) using the updated orientation of
the image. And the gray value for this point is
generated by resampling the input image. The DEM
can be the one derived from the same stereo pair or
it is digitised/derived from map. In case of map
DEM GCPs are also required additionally, to get
mapping between ground and image.
4.4 Quality Evaluation
One of the most critical component of the mapping
is the accuracy of the product used in terms of its
tickmarks and the internal distortion. The system
contains accuracy checks at every process level.
i.e. the model accuracy on GCPs and on check
points, is given immediately after space resection
and intersection. In addtion to this the Orthoimage
evaluation with respect to its tickmarks is done on
checkpoints within the system. The checkpoints are
identified on the Orthoimage manually, and their
estimated positions w.rt the tickmarks are
compared with the actual values. The RMS of these
errors are quoted as the location accuracy and the
standard deviation represents the internal distortion.
Apart from this quantitative approach, Orthoimage
is evaluated qualitatively by overlaying map features
either digitally or photographically. The DEM can be
qualitatively evaluated by (i) draping the
Orthoimage on the derived DEM and comparing this
with the draping of Orthoimage over map derived
DEM and (ii) comparing the contours obtained from
both image derived DEM and map derived DEM.
826
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996