International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004 
  
of rotating angle and translation value ( AZ ) in direction Z. 
As alternative, the error value was obtained indirectly from the 
mean shifting errors in the direction X and Y. on the other hand, 
the rotating angle and AZ are approximately zero because the 
object undergoes an approximate rectilineal moving in plane 
XY. This phenomenon is verified approximately by the 
experimental results of  . (0. K . Concluded from each 
step of the experiment, there are following aspects that take 
responsibilit of the error source between calculated values and 
true values in the process of estimating the entire movement 
parameters: 
(D The measuring precise of object space coordinates and 
corresponding image coordinates. The coefficient L, obtained 
from calculation has errors if the error exists in either object 
coordinates or corresponding image point coordinates because 
DLT reflects the translation relationship between object 
coordinate and corresponding image point coordinates. 
(2)The accuracy of stereo and sequence (moving ) matching is 
the crisis of the experiment. The error matching is inevitable in 
the process of matching. The strict correspondence of random 
feature points in object before and after is required in three- 
dimensional movement parameter that is based on point feature. 
In the present study, photogrammetry is combined effectively 
with computer vision. As a result, the methods and procedures 
is obtained to determine the parameters of object movement 
from the binocular stereo sequence images pair of moving 
object. Therefore, the solution of correspondence of three- 
dimensional feature point of moving object is obtained 
effectively. The future direction is to predict and estimate the 
moving object based on the current existing achievements in 
order to achieve the goal tracking determination. 
REFERENCES 
l.Z.Zhang, R.Derche, O.Faugeras and Q.T.Luong,A robust 
technique for matching two uncalibrated images through the 
recovery of the unknown epipolar geometry[J].Artificial 
Intelligence 1995,78(1-2): 87-119. 
2.Aggarwal J K,Nandhakummar N.On the computation of 
motion from sequences of images: a review. In Proc.IEEE, 
1988,76(8): 917-935. 
3.Xu Wenli, Liu Wenhuang Estimation motion parameters from 
position vectors of feature points. ACTA Automatica Sinica 
Vol.18.No.4, July, 1992:440-447 
Project supported by the National Science Foundation of China 
(No.40171080) 
     
    
    
   
    
   
   
  
     
    
     
      
   
    
    
KE 
ABS 
We 
sequ 
desi 
estir 
pixe 
sim] 
Test 
pro 
prec