anbul 2004 
the typical 
d involves 
a template 
I) with its 
mages. 
| Searching 
ün a new 
^, Stop the 
iterations, 
tels. If the 
divergent 
id point". 
wall. Their 
'€ 4 shows 
alid points 
ectively. It 
r trackable 
mentioned 
[ trackable 
ntinuously 
image 1 to 
oes in the 
number of 
pixels still 
ge number 
nce their 
mage shift 
on on the 
S. the LK 
th shorter 
egistration 
is used as 
egistration 
df error- 
FE method 
n than the 
set of test 
egistration 
nost time- 
r the tests 
» maximal 
the mask 
lows some 
. the IOFE 
. Figure 8 
ethod can 
frames per 
  
Figure 3. A digital video image of near 2D objects on a wall 
number of vectors 
(A: tracking from image 1 to 2) 
       
  
            
        
         
D 5 4D 145 20 25 30 35 40 45 
displacement vector length (pixels) 
number of vectors (B: tracking from image 1 to 3) 
2001]-——— i nde 4 SH AEE ale bes ces eee 
160 p. 
140 
  
  
  
  
  
120 | 
100 
A 
————— 
t 
  
  
  
  
  
  
  
  
  
  
mee pe ns T 
OQ 5. 4D 15 20 25 30 35 40.45 
displacement vector length (pixels) 
  
(C: tracking from image 1 to 4) 
  
number of vectors 
200 
180 
1601 
140 
   
  
  
  
  
  
  
  
  
a TS dmm errare eder 
à 5 10 15 20 25 20 35 40 45 
displacement vector length (pixels) 
  
  
  
  
number of vectors 
160ft- 
140 
100] ........ P p : ront: : e— seit mec tee d do 5 
(D: tracking from image 1 to 5) 
  
  
  
  
  
a 
Ta 
0. 3 0 4 2 25 N s 4 «4 
displacement vector length (pixels) 
  
  
  
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
number of vectors (E: tracking from image 1 to 6) 
140]. 
120 
100} 
80} 
60 
40 
20 
  
    
   
   
  
  
  
  
0 : : : : : : 
0 5 10 15 20 25 3) 35 4 4 
displacement vector length (pixels) 
number of vectors 
160 
140 
120 
1007 
80 
60 
40 ; ; : : : ; : ; 
ol : i ; : : : 
0 5 10 15 20 25 30 35 
displacement vector length (pixels) 
(F: tracking from image 1 to 7) 
  
  
   
  
  
      
  
40 45 
Figure 4. The IOFE method for tracking from the image 1 to 2 (A), 
from 1 to 3(B), from 1 to 4 (C), from 1 to 5 (D), from 1 to 6 (E), 
and from 1 to 7 (F) (number of extracted feature points = 650) 
number of vectors 
ü 14 2 38 4 5 67 88 101111202 13 id 
displacement vector length (pixels) 
number of vectars 
ol ten 
250 
2001 
Apdo ied ria 
100 E es iu j : | : : : | | | 
50 Siri a Ra vu 
Eine be 
0 12 8455 780891612134 
displacement vector length (pixels) 
  
  
  
  
  
  
  
  
Figure 5. Histograms of displacement vector lengths: the LK 
method (top) and the IOFE method (bottom), where light-blue and 
dark-blue lines denote the ones without and with error detection