point in right image. 
3) finding the matching point in left image 
using the patch as target image. 
4) differencing the positions of the two left 
points. 
5) if the difference does not exceed the given 
value the system will accept the matching result. 
The QC method has been found to be effective. Soon 
the AEDAS will use a new method to detect and remove 
the gross error. The following is the details: 
1) orthorectifying with left and right image 
respectively. That means the system gets wo 
orthophotoes that shoudn't have any deformation and 
that only have gray difference. 
2) matching the orthophoto in desired grid. In 
the theory, the matched point is the same point. If 
the ofsset does not equal zero, the orthophoto patch 
around the point and the height of the point is 
wrong. However, it doesn't matter because the offset 
is just the corection. So the system can correct the 
height and re-orthorectify the patch. 
Outputs 
The outputs consist of precise object space 
positions, orientation data, DTM, orthphoto data and 
every kind of QC data. 
EXAMPLE AND RESULTS 
In the example a pair of photo will be 
processed. The basic information of the data source 
is listed as follows: 
photo size 23 x 13 cm 
photo scale 1: 10000 
scanning resolution 50 microns 
image size 4600 x 4600 pixels 
DIM interval 20 m 
The following is the main production time and 
accuracy: 
Inner orientation 2/30" 
Relative orientation 9^30" 0.0045 mm 
Resamp le 10700" 
DTM Generation 15700" 0.2 pixel 
(150x120 points) € 20 points per second) 
DTM editing 37.00” 
Ortho 10/00” 
CONCLUSION 
The AEDAS system has been found effective by 
experiments. In the future, we wil perfect our 
system by adding other functional part to AEDAS such 
as photo scanner for scanning, mosaicing to 
mapsheet, outputing to film recorder, perspective 
scene generation, even better matching method. we 
also will make use of the VÀ-1 vector accelerator to 
save processing time greatly. 
272 
REFERENCES 
[1]. G.Borgefors, "Hierarchical chamfer matching: 
a parametric edge matching algorithm" IEEE PAMI Vol. 
10. No. 6. pp. 849-865, 1988. 
[2]. S.T. BARNARD and W.B. Thompson  "Disparity 
analysis of images" IEEE PAMI. Vol.2, N0.4 pp. 
333-340. 1980. 
[3]. W. Forstner and E. Gulch. "A fast operator 
fordetection and precise location of distinct point, 
corners and centres of circular features” ISPRS 
Intercommission Workshop, Interlaken, 1987. 
[4]. C. W. Molander and J.F. Otto, " Digital 
terrainmodel data extraction and editing on the 
pegasussoftcopy workstation” Technical Papers 
ACSM-ASPRSAnnual Convention, Baltimore, 2: pp. 
151-161 1989. 
[5]. M.Claus, " Digital Terrain Models through 
Digital Stereo Correlation” , Pattern Recognition in 
Photogrammetry, Graz 1983. 
re 
wit 
KE 
di