International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
Fig. 5 shows an ORIMA stereo point measurement interface 
and graphical point and image display. 
5 AUTOMATIC DTM EXTRACTION 
Automatic generation of terrain models from overlapped 
images is currently the most important way to collect a DTM. 
The DTM extraction module in LPS concentrates on improving 
the reliability, speed and productivity of the DTM extraction 
process. In order to improve reliability and speed, an integrated 
image matching strategy is developed. It includes a feature- 
supported image correlation, geometric constraints, hierarchical 
point search, on-the-fly resampling for rotation and distortion 
compensation, consistency checking etc. For mproving speed 
and productivity, the DTM module can generate DTM’s for all 
overlapped pairs in the block with the option to automatically 
mosaic them into a single optimized output. For further 
flexibility, some easy-to-use graphical interface and tools are 
developed, with which a user can manipulate their image pairs 
and different interest areas such as excluding regions and 
Strategy parameters very easily and efficiently. It supports 
variety of input image formats and many output DTM types 
such as TIN, raster DEM, ESRI 3D shapefile and ASCII. 
Furthermore, The LPS DTM extraction module works for 
[em 
         
Figure 6a: Graphic overview of the block layout with 
GCPs, check points and automatically collected tie points 
  
Figure 6b: Automatically extracted DEM for the whole 
block with grid size 1x1 meters without any editing 
all aerial images, digital camera images, video camera images 
and satellite images, which offers excellent economy for the 
customers. Fig. 6 shows an example of a 28-image block used 
in the OEEPE test. It covers an area of the town Forssa in 
Southern Finland. The image scale is 1:4000 and flight height is 
about 600 meters. The image overlap in strip is 60% and side 
overlap varies from 24% to 49%. The film is positive color and 
is scanned as black and white. The pixel size is 30 micrometers: 
each scanned image has 8000x8000 pixels. There are 14 known 
840 
ground points available for the test. LPS APM works well using 
the provided approximate coordinates of image photographic 
center. It generates 437 object points with 1477 image points. 
The triangulation computation includes automatic gross error 
detection and 4 additional parameters. 8 of the 14 known 
ground points are set to be GCP's and the remaining 6 are used 
as check points. The root mean square error of 6 check points is 
0.041, 0.054 and 0.055 meters for X, Y, Z respectively. The 
Figure 6a shows the graphic display of the triangulation results. 
It is the footprint of the block with ground point positions and 
residual vectors. The triangle symbol represents GCP's, the 
circle symbol represents check points, and the rectangle symbol 
represents the automatically collected tie points. (Since the 
residuals of GCP's and check points are very small, they are 
difficult to see on the reduced graphic.) Considering the 
realistic conditions of the block, the accuracy of the 
triangulation results looks very reasonable. Figure 6b shows the 
automatically extracted DEM for the whole 28 images with 
ground cell size 1x1 meters. The whole automatic DEM 
extraction computation takes less than 30 minutes for 28 
images on an ordinary Pentium 4 with 2.4GHz Laptop PC. 
Using 383 triangulated tie points (excluding the tie points 
outside DEM boundary and on buildings and trees) as ground 
check points, the automatic DEM has accuracy about 0.2 
meters. The result shows the automatic DEM process is very 
fast and accurate. 
6 SUMMARY \ 
This paper has introduced a new digital photogrammetric 
workstation, namely Leica Photogrammetry Suite™. Its main 
functional modules and system workflow are described. The 
features and algorithms for sensor modeling, automatic interior 
orientation, automatic point measurement, block triangulation 
and automatic DTM extraction of different kinds of images are 
introduced. Several examples are demonstrated. It shows a 
powerful, production oriented, process driven and easy to use 
digital photogrammetric workstation is ready to use for both 
advanced photogrammetrists and GIS professionals to 
accurately and efficiently convert their geospatial imagery into 
usable geospatial data. 
ACKNOWLEDGMENTS AND REFERNENCES 
To be added. 
lii 
fi