All photogrammetric basic tasks like exterior, relative 
and absolute orientation as well as space intersection 
can be resolved with feature based methods. But like 
in pointwise approach also in feature based procedure, 
the distribution of control features is essential for 
reliable results’. 
In modeling from video frames we use feature based 
triangulation to solve the modeling problem. After 
having a successful feature matching, the operator can 
point out control features for the system. Also similar 
kind of feature matching as in Chapter 4 can be 
applied for detecting control feature and 2D feature in 
image space correspondence. Before the final bundle 
adjustment, all those feature lines have to be rejected 
from sample space which are parallel or nearly 
parallel with the image base line. An exception is 
when those lines have a predetermined intersection 
between lines with different direction properties. 
Otherwise, this would lead into a singularity problem 
in the estimation. Also the use of parallel imaging 
strips like in aerial photography will eliminate this 
problem. 
6. SUMMARY 
The proposed algorithm combines automatic ob- 
servation extraction, robust feature matching and use 
of linear features in modeling. The idea has been to 
utilize the power of LSQ-estimation by increasing the 
redundacy . This has been done in two ways; using 
massive number of frames in observation extraction 
phase and including all edge points of an object 
feature to determine its parameters. 
The automation has not been carried out through the 
whole process, but only on most computing consuming 
phases. The operator inspection and assistance will be 
used in the process to guarantee the convergence of 
the system. 
7. REFERENCES 
  
' Mulawa, D.C., Mikhail, E.M., 1988. Photogrammetric 
Treatment of Linear Features. In: Int. Arch. 
Photogrammetry and Remote Sensing, Kyoto, Japan, 
Commission III. 
? Mulawa, D.C., 1989. Estimation and 
photogrammetric treatment of linear features. UMI 
Dissertation information service, Purdue, p.312. 
* Rosenfeld, A., Kak, A., 1982. Digital picture 
processing. Vol. 1-2, Computer Science and Applied 
Mathematics, Academic Press, Orlando, 2nd edition. 
“ Canny, J., 1986. A Computational Approach to Edge 
Detection. IEEE Trans. on PAMI. Vol. PAMI-8, no. 6, 
pp. 679-698. 
224 
  
° Illingworth, J., Kittler, J., 1987. The Adaptive Hough 
Transform. IEEE Trans. on PAMI. Vol. PAMI-9, no. 5, 
pp. 690-698. 
° Princen, J., Yuen, H.K,, Illingworth, J., Kittler, J., 
1989. Properties of the Adaptive Hough Transform. 
Proc. of 6th Scandinavian conference on Image 
Analysis, Oulu Finland, pp. 613-620. 
7 Princen, J., Yuen, H.K,, Illingworth, J., Kittler, J., 
1989. A Comparison of Hough Transform Methods. 
Proc. of IEEE 3rd International Conference on Image 
Processing and its Applications, University of 
Warwick, pp. 73-77. 
* Xu, L., Oja, E., 1993. Randomized Hough Transform 
(RHT): Basic mechanisms, algorithms, and compu- 
tational complexities. CVGIP: Image understanding, 
vol. 57, no. 2, pp 131-154. 
? Heikkinen, J., 1994. Linear Feature Based Approach 
to Map Revision. Int. Arch. of Photogrammetry and 
Remote Sensing. Athens, Georgia, U.S.A. Commission 
IV Symposium, pp. 344-351. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996 
KEY WORD 
Geometry, ^ 
ABSTRACT 
This paper 
structures. ' 
either a Per: 
SEM image 
ment, which 
have very d 
relevant ime 
processing ! 
feature-bas« 
The three-c 
range plays 
or material 
quality con 
and easy-tc 
ments. 
The combir 
image acqu 
(SEM) offe 
an exceller 
wide range 
image proc 
3D-measur 
The devel 
takes spec 
into consi 
changing ir 
possibility c 
to increase 
mination. 
A further a 
methods ii 
techniques 
different su 
to achieve 
face geom 
approache 
nearly con 
Digital Ele 
ching metr 
On the oth 
microstruc