The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
422
were utilized to do adjustment of camera model as GCPs. Under
ordinary circumstances, both old and new features were
extracted from the old aerial photos and PRISM’s ortho images.
Figure 10. Linear Features and Common Points (Area 4)
The reference of linear features is weighted for small change.
This is more easily obtainable and more easy-to-use values than
other references. Therefore huge development of artificiality
has a tendency to trace a part of original feature in Japan.
5. EFFICIENT ORTHO-RECTIFICATION
Finally, based on the above result, we tried a streamlined
workflow of ortho-rectification. A result of accuracy validation
is shown in Table 2.
X
Y
TOTAL
Area 1
1.5439
1.4306
2.1048
Area 4
4.6497
0.5172
4.6783
Table 2. Check point error
Figure 12. GCPs and check points (Area 4)
6. CONCLUSION
The mentioned results lead to the conclusion that the use of
PRISM products is effective in ortho rectification of old aerial
photos, especially for the medium and small cities. However,
some rectifications of DTMs are essential around developed
hilltops and damaged steep slopes or gullies. For such problems,
we will have some rectifications of the terrain model with
experimental filtering in the near future. Moreover, full-
automation of collection of GPSs also remains. Especially in
this case we could not get sufficient information on camera
parameters and the flight courses. To utilize the wealth of
image archives, we must try to add a feature matching
technique that can correspond to rectify the skew of extracted
linear features.
REFERENCES
Geographical Survey Institute, Government of Japan, 1995.
GIS HOME PAGE, Japan. http://www.gsi.go.jp/ENGLISH/
(accessed 30 Apr 2008)
Remote Sensing Technology Center of JAPAN (RESTEC),
2000. RESTEC, Japan, http://www.restec.or.jp/top_e.html
(accessed 30 Apr 2008)
Figure 11. Workflow with ERDAS IMAGINE
