inbul 2004 
    
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
  
d by using base-plane. 
culated by 
(12) 
  
33 
-1 
(t+1) 
among 
* 
"T 
"T 
wa.» 
. 
  
lane and a 
  
5. EXPERIMENTAL RESULTS 
  
Fig. 8. Multi-baselines calculated by using an image sequence. 
  
(a) Vectors of multi-baselines cut useless parts 
  
  
    
(d) View-port with down ward angle 
Fig. 9. Projection of image frames to multi-baselines. 
For realizing the proposed method and displaying result in 3D 
space, we use Visual C++ 6.0 and OpenGL. The propose 
method has been applied to real image sequence taken from a 
translating and rotating video camera. Fig. 8 is about the vectors 
of multi-baselines in connection with the whole size of image 
frame. Fig. 9(a) is about cutting of useless part of the multi- 
baselines in Fig. 8 and Fig. 9(b) shows the left side of a house 
and Fig 9(c) shows the front side of the house in the result 
created by projecting image frames on the multi-baseline in Fig. 
9(a). Figure 9(d) shows the result at view-port with downward 
angle. 
6. CONCLUSION 
The novel method for creating video mosaics in 3D space 
based on image sequence taken from a translating and rotating 
camera has been proposed in this paper. The core of the 
proposed algorithm is to create multi-baselines of which each 
baseline is determined by using the pose of camera and the 
average distance between camera focus and depths of optical 
flows per image frame. The proposed method is successtully 
applied to real-image sequence taken from a translating and 
rotating camera. 
In case that the proposed method is applied to the video 
camera of vehicle-borne mobile mapping system which 
generally includes rotational and translational motion, the 
results by the proposed method will be applied to extract 
textures of facades and obtain rapid and inexpensive 3D spatial 
data instead of laser scanner. Also, since the results are 
continuous data, the results will be overcome the drawback of 
panoramas and spherical mosaics sequence of which the data is 
discrete for visual environment in large area. 
References from Journals: 
Assaf Zomet, Doron Feldman, Shmuel Peleg, and Daphna 
Weinshall, * Mosaicing New Views: The Crossed-Slits 
Priojection," IEEE Transactions on PAMI, Vol. 25, No. 6, June 
2003. 
Coorg, S. and S. Teller, 2000. Spherical mosaics with 
quarterings and dense correlation, /n JJCV, 37(3): pp. 259-273. 
Shum, H. and R. Szeliski, 2000. Systems and experiment paper 
construction of panoramic image mosaics with global and 
local alignment, /n JJCV, 36(2): 101-130. 
Smith, S.M. and J.M. Brady, 1997. SUSAN-a new approach to 
low level image processing, /n JJCV, 23(1): 45-78. 
Szeliski, R, 1966, Video mosaic for virtual enviroment, Comput. 
Graph. Applicat., vol.16, no.3, pp. 22-30. 
References from Other Literature: 
Chou, J. S.,and J. Qian, Z. Wu, H. Schramm, 1997. Automatic 
mosaic and display from a sequence of peripheral angiographic 
images, Proc. of SPIE, Medical Imaging, California, 3034: 
1077-1087. 
Coorg, S. and S. Teller, 1999. Extracting textured vertical 
facades from controlled close-range imagery, Proc. of 1999 
IEEE Conference on Computer Vision and Pattern Recognition, 
Colorado, pp.625-632. 
Zhu, Z., A. R. Hanson, H. Schultz, E. M. Riseman, 2001. Error 
Characterics of Parallel-Perspective Stereo Mosaics, Proc. of 
IEEE Workshop on Video Registration (with ICCVOI), 
Vancouver, Canada, 13 July. 
References from websites: 
USGS Hurricane Mitch Program Projects, 
http://mitchnts 1 .cr.usgs.gov/projects/aerial.html