air by the 
of balloon 
50m. Also, 
e at the 
rns; one is 
urning and 
er is to turn 
ime speed. 
ormal size 
vehicle. 
ed and 
of running 
'asurement 
ie following 
f cars, 2) 
arison by 
he kind of 
ilculated by 
in order to 
time. The: 
type of car 
ition of car 
the type of 
/een actual 
ideo image 
  
  
taken at 50m high above the ground. 
The car of black body showed larger measuring error 
and larger fluctuation in measured value comparing with 
other kinds of car. This is because tracing of car by 
matching was incomplete in many cases. When the 
body color is black, it was difficult to distinguish the car 
from the asphalt surface of road having the similar color, 
and matching tended to be incomplete. 
(2) Comparison of measuring results by running 
speed 
Figure 4 shows the measuring results of white car as 
an example of comparison by running speed. 
The measuring results are roughly classified into two 
groups. One is the group which has small deviation from 
the actual measured value and the other is the group 
having the larger deviation. The difference of these 
groups has resulted from the successful or unsuccessful 
matching in image correlation. In case of the group 
having larger deviation, there was a rainfall during the 
test run, and there were puddles and speckled patterns 
on the road. As a result, matching was not well done 
because it was difficult to distinguish the car body from 
road surface. 
In case of the group with successful matching, large 
difference was not observed in the measured value from 
the actual value even when the running speed was 
different. The same can be said to the large sized 
vehicle. Seeing from the spatial resolution of video 
camera and photographing altitude, the range of 
measured value had been considered to be about 5 
km/h of the actual value. Among the measuring results, 
those with successful matching showed a result almost 
consistent with this expectation. 
(3) Comparison by observation altitude 
Assuming the cases where the balloon is sent up 
around the intersection, we compared the measuring 
accuracy by different observation altitudes of 50m, 
100m and 150m. In this experiment , it was possible to 
cover the same area at any altitude by adjusting zoom 
lens. Therefore, hardly any difference was observed in 
the measuring result when the difference of altitude is 
within the extent of 50m -150m. 
Based on the discussion made in the above, the 
methods of taking video image and measurement used 
in this study are considered to provide the measuring 
values which are close to the accuracy logically 
determined by the performance of CCD camera and 
photographing altitude only if matching under the image 
correlation is in good condition. 
5. Measuring Results of Running Locus and 
Discussion 
In the discussion on measuring result of running 
locus, we verified the measurement accuracy when the 
loci of vehicles turning at three different radii of bm, 10m 
and 15m were measured. The viewpoints of discussion 
were (1)appropriateness of measuring results, (2) 
comparison of the results taken at different observation 
altitude and (3) evaluation of the influence of mid-way 
stop at turning. 
(1) Appropriateness of measuring results 
  
  
  
70 : 
SMELL | MM Pr desk 
E [P e MS e Cd à ndun: iisontis at td a. ae hand cniin 
x RT TT 
o : | | | 9 e : 
E msew iS on NN a ie nm jenem. reete pias vues: haa CHEER 
S { ; ; : : ; ; 
UM C und EE DET E em iege 17 0 0 SUITE pe T 
3 | : | | | | | 
Bf gf git 
= : | 999 : : | | : | | | | : 
10 [oo er E UO $i obrera He eem 4 etie det 
0 A A à f f f a s 2 [1 s A 
35 40 45 50 55 60 65 70 
Actual value(Km/h) 
Figure 4 Actual running speed and measured value 
(in case of photographing altitude 50m, white normal size car) 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996