Figure 5: Four different objects with extracted breaklines. 
dataset and mapped them to 3D using the registration 
found in the previous step. As can be seen from the overlay 
in Fig. 4(f), combining the intensity image extraction and the 
range sensor extraction gives the breaklines, as desired. 
Fig. 5 shows the result of this extraction for four different 
objects. It can be seen that the upper right object can be 
distinguished immediately from the others. Thus, we have 
performed one cycle in our object recognition system of 
Fig. 3. To distinguish the remaining parts, e.g. the shape 
can be used. From Fig. 5 we see that the upper two parts 
are shaped oylindrically, while the lower two parts contain 
a large plain region. Capturing these differences can be 
done e.g. by using a camera in connection with dedicated 
light incidence angles. 
5 OUTLOOK 
We have proposed a new object recognition concept that 
incorporates multi-sensor fusion and active exploration into 
the recognition process. From multi-sensor fusion, we ex- 
pect to obtain a significant quality improvement of the seg- 
mentation result. Active exploration is a key feature to avoid 
the problem of combinatorial explosion of the interpretation 
tree. We have shown some first examples which illustrate 
our proposed concept. 
ACKNOWLEDGEMENT 
We thank Mr. C. Voland of the Institute for Technical Optics 
for providing the range images. 
64 
References 
Bhanu, B. & Ho, C.-C. (1987), ‘CAD-based 3D object repre- 
sentation for robot vision’, IEEE Computer 20(8), 19— 
35. 
Bolles, R. C., Horaud, P. & Hannah, M. J. (1983), 3DPO: 
A three-dimensional part orientation system, in ‘Proc. 
of the Ninth International Joint Conf. on Artificial Intel- 
ligence, Karlsruhe, Germany’, pp. 1116-1120. 
Eggert, D. W., Bowyer, K. W. & Dyer, C. R. (1992), Aspect 
graphs: State-of-the-art and applications in digital pho- 
togrammetry, in ‘Proc. ISPRS Congress 1992, IAPRS 
Vol. 29 Part B5 Comm. V’, pp. 633—645. 
Flynn, P. J. & Jain, A. K. (1991), ‘CAD-based computer vi- 
sion: From CAD models to relational graphs’, IEEE 
Trans. on Pattern Analysis and Machine Intelligence 
13(2), 114—132. 
Grimson, W. E. L. (1990), Object recognition by computer, 
Series in Artificial Intelligence, MIT Press, Cambridge, 
Massachusetts. 
Grün, A. (1994), Digital close-range photogrammetry — 
progress through automation, in 'Proc. Comm. V 
Symp. Close Range Techniques and Machine Vision, 
IAPRS, vol. 30, part 5', pp. 122-135. 
Hansen, C. & Henderson, T. C. (1989), 'CAGD-based com- 
puter vision’, IEEE Trans. on Pattern Analysis and Ma- 
chine Intelligence 11(11), 1181-1193. 
ISO (1994), ISO 103031 Product Data Representation and 
Exchange Part 1: Overview and Fundamental Princi- 
ples. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996 
KEY WOR 
Two photo, 
rockfill dar 
stability of 
and target: 
photocoord 
determinat: 
1. Introdu 
Since 197; 
the rockfill 
photogram: 
method - 
Congress I 
Optical 3] 
(Cernansky 
The SAS 
photogram: 
times per 
angles me: 
on the pl 
projection 
eccentricit; 
For the sp: 
also the C 
1991). It 
determinec 
ground sur 
2. Detern 
intersectic 
For the di: 
network w 
was estab 
around the 
theodolite 
The detail 
All these