y accurate. 
might be 
algorithm 
jants once 
3 computed 
absolute or 
coefficients 
coefficients 
have been 
mputation 
hosen from 
t belong to 
the second 
] matching 
entification 
not known 
first basic 
; to random 
that have 
|, 16%, 25% 
cond curve 
curve cases 
sfficients of 
area than 
est 8 cases 
ad consecu- 
| generated 
| of the 16 
X tion under 
s two shifts 
aph has 48 
, second 16 
ars are for 
h the mean 
ro. If there 
rmation is 
os four bars 
Ast squares 
ation after 
were dot 
ectively. It 
mputations 
formation. 
he method 
% and 25% 
m the first 
inal motion 
That was 
» invariants 
Differences added to random locations. Degree of a curve is 3 
  
I 
  
thon Orne [I T im © | 
  
Ih [Hi 
  
16% 25% 3596 50% 
Differences added to constant and consecutive positions. Degree of a curve is 3 
  
  
  
  
j m T 
  
  
non [ho Iri lil 
  
m4 Li] T 
  
Differences added to random locations. Degree of a curve is 1 
  
| 
  
  
  
  
  
  
  
  
me (oan hn 1 D Li 1 nl] il I 
  
Differences added to constant and consecutive positions. Degree of a curve is 1 
  
  
rs mean Tit [TT] D br II 
  
Figure 5 
  
  
  
  
  
  
  
mani I 
16% 25% 35% 50% 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996