tion, 
) 
ti-layer) 
ation 
  
that all sensor data 
ation. In our case, 
icted object surface, 
This requires the re- 
p is done using the 
! measuring system, 
expensive position- 
to use given sensor 
ind to fit the data ac- 
in be identified auto- 
strate this approach 
ind location concept, 
its. Fig. 4(a) shows 
camera of the stereo 
-form shaped sheet 
the metal has been 
have a resolution of 
oarse height model 
xpected, this coarse 
yreaklines of the cut- 
SS, since the cut-out 
sity imagery, we can 
ocessing. As shown 
5 some spurious data 
we use the larger of 
of interest which are 
lower left part of the 
256 3D data points. 
es very well, the data 
a 1996 
  
Figure 4: (a) original image of the object. (b) coarse resolution height model obtained by image matching. (c) result of 
intensity image feature extraction, transformed into 3D space. (d) range data set. (e) result of fitting range data set to the 
height model in (b). (f) overlay of features extracted from intensity and range images. 
is also suited to assess the surface roughness. Since image matching to register both datasets and obtain the re- 
we have for both the photogrammetric and the range data sult depicted in Fig. 4(e). 
height model synchronous intensity information, we used 
In a further step, we extracted features from the range 
63 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996