tanbul 2004
he test field
are obtained
uracy of the
The vertical
map of GSI,
Iso used to
:al residuals
ry (Fig. 12)
€ as above.
is shown in
e middle of
ange of this
SMs is 5m.
' horizontal
> of vertical
s is 10.1m,
m of that is
s.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
50000
|
|
60000 TE |
|
40000 |
numberofponts
30000 Lil
20000
10000 Re erum s A eru ia A
0 i hho ad 000 00 00 000 00 0084
VO VV VV VV VV VV VPI vv oY
uy oo uuoumo utcoium oie toting ro
GQ 0) 145 0 d oO N Nx DS
e ei r1 v ci
difference [m ]
Fig. 14 the comparison between two DSMs
About 6 hundred points (0.5% of whole points) have the
difference of more than 100m.
4. EVALUATION Of Orthorectified Image
We generated DSM from QuickBird stereo imagery of
Yokosuka City and its surroundings automatically. The grid
interval of DSM is also Sm. Then we orthorectified QuickBird
sterco imagery by using this DSM and obtained two
orthoimages, one is from the forward view image (Fig. 15) and
the other is from the backward view image.
FS
€2004DigitalGlobe
Fig. 15 orthoimage from QuickBird imagery
In order to evaluate the orthoimages, we measured the
horizontal coordinates of 57 check points from two images.
Then we compared the coordinates with each other and the
result of the GPS survey. The result is shown in Table.1. In the
table “Forward”, ”Backward” and “GPS” mean the coordinates
measured from an orthoimage of the forward view image, those
of the backward view image and the GPS survey results
respectively. The difference is absolute value in meter.
Table.1 the comparison between horizontal coordinates
Unit : m
Differece Forward GPS GPS GPS
etween: and and and and
of Backward | Forward | Backward Average of
Differece Backward
and Forward
Maximum 50.6 27.7 27.1 4.1
RMS 11.5 6.2 5.6 1.4
Average 3.3 2.7 2.2 1.1
Maximum in
northing 49.5 27.0 27.0 2.6
RMS in
northing 11.3 59 5.6 1.1
Average in
northing 3:7 2.5 2.1 0.8
Maximum in
easting 10.4 8.2 2.2 3.1
RMS in
easting 2.2 1.9 0.7 0.8
Average in
easting 0.9 0.8 0.5 0.5
The differences in northing are much larger than those in
easting. The offnadir angle of QuickBird imagery is large in the
north-south direction, then the software probably sometimes
cannot find true matching points for DSM generation along the
north-south direction and the images cannot be corrected
appropriately.
Accurate coordinates were obtained by calculating the average
of coordinates measured from the orthoimages. However, these
orthoimages are not applicable for topographical mapping.
5. Conclusion
In this study the DSM and the orthoimages from QuickBird
imagery are evaluated. The 3D model from QuickBird imagery
is accurate. However, the DSM contains the large error and the
orthoimage is not satisfied for topographical mapping.
QuickBird imagery has high potential, and if we create DSM
from QuickBird imagery manually, the DSM and the
orthoimages will be accurate. The result shows the status of the
automatic generation of the DSM and orthoimages using
satellite imagery.
Acknowledgements
Hitachi Software Engineering Co., Ltd. offered the images
evaluated in this paper. We express our appreciation.
747
