control lines by combining two points such as 1-18, 2-17, 
3-16, 4-15, and so on. In addition, four pairs of parallel 
lines were used here. Table 6 listed the tested result 
obtained with the new method. The interior parameters 
were provided by the former two methods. 
Table 6. Result of Single Photo Resection 
  
  
  
  
  
Initial Value Calculated Value 
X, 83.296 -- 
Y, -57.371 vee 
f 239.396 --- 
Xs 5300.991 5366.513 
Ys 900.396 966.624 
Zs 30.896 37.255 
9 110.601898 115.577747 
© 90.394714 91.347435 
K 0.140747 359.280641 
  
5. CONCLUSION 
Calibration of camera has been an important 
component of any vision task which seeks to extract 
geometric information from a scene. The new method 
presented in this paper to determine the focal length 
and the exterior orientation parameters of camera is 
based on straight lines and their geometric constrains. 
The straight lines were used because they could provide 
excellent calibration environment for the object controls. 
Compared with other methods the new technique 
provided here possesses such advantages: 
€ (Calibration of camera, which up to now has been 
point based, can be implemented on the basis of 
linear features. The equations then relate feature 
descriptors instead of point coordinates. 
€ Linear features, which are abundant in the 3D 
world due to human-made infrastructure, provide 
a rich set of possible geometric constraints (e.g. 
parallel, perpendicular, horizontal, vertical and 
coplanar) that can be effectively exploited in 
different applications, particularly in 
photogrametry. 
€ Although some techniques based on linear 
features have been put forward (Mikhail, 1997; 
Echigo, 1990; Grosky et al, 1990; Wang et al, 
1991; Chen et al., 1989; Chen et al., 1991; Lee et 
al, 1990; Lenz et al, 1988; Liu et al, 1990; 
Mulawa et al, 1988; Salar et al, 1990; 
Tommaselli et al., 1988; Tozzi, 1996; ), few of them 
took advantage of these excellent geometric 
constrains. The new method makes full use these 
constrains (mainly vertical parallel and horizontal 
lines). This can be proven both in geometric theory 
and mathematical basis. 
€ Since the calibration was divided into two steps, 
this can obviously reduce the relativity among the 
orientation parameters. 
€ The new method is robust and stable due to its 
strong geometric and mathematical relations. 
€ By the technique the calibration of camera can be 
done automatically because the straight lines can 
be easily extracted from the digital images at the 
sub-pixel accuracy. 
The proposed solution was tested using synthetic 
28 
and real data. From table 6 we can see that the new 
technique could obtain the same accuracy level result 
as DLT and point-based space resection. So it could be 
appropriate both for outdoor and for indoor computer 
vision applications like robot location and autonomous 
land vehicle guidance, because of its simplicity of 
environment setup and strong geometric controls. 
Especially in the application of the merging image with 
existing 3D GIS models the new method might be the 
most suitable way to provide the exterior orientation 
parameters because it is easy to provide more line 
objects than point from the GIS database. 
6. REFERENCES 
CHEN, S.Y. and TSAI, W.H., 1991, Determination of 
Robot Location by Common Object Shapes, IEEE 
Transactions on Robotics and Automation, Vol. 7, No. 1, 
pp. 149-156. 
CHEN, Z.; TSENG, D. and LIN, J., 1989. A Simple 
Vision Algorithm for 3-d Position Determination Using 
a Single Calibration Object, Pattern Recognition, Vol. 
22, No. 2, pp. 173-187. 
CHEN, W. and JIANG, B. C., 1991. 3-D Camera 
Calibration Using Vanishing Point Concept, Pattern 
Recognition, Vol. 24, No. 1, pp. 57-67. 
ECHIGO, T., 1990. A Camera Calibration Technique 
Using Three Sets of Parallel Lines, Machine Vision and 
Applications, Vol. 3, pp. 159-167. 
LENZ, R. K. and TSAI, R. Y.,, 1988. Techniques for 
Calibrations of the Scale Factor and Image Center for 
High Accuracy 3-D Machine Vision Metrology, IEEE 
Transactions on PAMI. Vol. 10, No. 5. 
MIKHAIL, E. M. and WEERAWONG, K. 1994. 
Feature-based Photogrammetric Object Construction, 
Proc, Am. Soc. For Photogrammetry and Remote 
Sensing Annu. Convention, ASPRS, Bethesda, Md., I, 
pp. 408-417. 
MULAWA, D. C. and MIKHAIL, E. M, 1988. 
Photogrammetric Treatment of Linear Features, In: 
Proceedings of the International Society for 
Photogrammetry and Remote Sensing, 1988, Kyoto, 
Commission III, pp. 383-393. 
TOMMASELLI, A.M.G. and LUGNANI,J. B., 1988. A 
Filtering Based Approach to Eye-in-Hand Robot Vision, 
In: Proceedings of the International Society for 
Photogrammetry and Remote Sensing, 1992, 
Washington, Commission V. 
I. Fukui, 1981. TV image processing to determine the 
position of a robot vehicle, Pattern Recognition, vol. 14, 
pp. 101-109. 
H.L. Chou and W.H. Tsai, 1986. A new approach to 
robot location by house corners, Pattern Recognition, 
vol. 19, no. 6, pp. 439-451. 
S.P. Lio and R.C. Jain, 1987. Road following using 
vanishing points, Computer Vision, Graphics, Image 
Processing, Vol. 39, pp. 116-130. 
R.Y. Tsai 1987. A Versatile Camera Calibration 
Technique for High Accuracy 3D Machine Vision 
Metrology Using Off-the-Shelf TV Cameras and Lenses, 
IEEE J. Robotics and Automation, vol. 3, no. 4, pp. 323- 
331. 
KEY WORD! 
ABSTRACT 
The move fro 
rapidly. This i 
photogrammet 
over large are 
environments, 
This paper des 
Each node con 
signal processc 
computer for 3 
rate of over 17 
A programme 
measurement s 
initial results v 
al, 1996(a & 1 
1996;) concer 
presented (Cle 
ongoing devel 
networked inte 
2.1 Hardware 
The 2-D proce 
ADSP-21xx fa 
of a DSP mod 
video feature 
communication 
(PSU-90) modi 
system where 
and sub-pixel 
(Figure 1). 
The VFE-90 
analogue and 
hardware leve 
considerably. 
photogrammetr 
VFE-90 modu 
converted into 
is stored in a : 
object above | 
location is plac