The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3b. Beijing 2008 
576 
Figure 1. The relationship and structure of topics in this paper 
2. INTEGRATION OF MULTI SPECTRAL RS 
IMAGERY DATA WITH FCD. 
No matter in the countryside or in urban areas, the traffic state 
on the road network can be surely monitored by direct 
measurements (e.g. induction loops and radar device). This 
traditional method is effective only when there are not so many 
vehicles and the demand on the monitor is low, because those 
devices could not, to some degree, monitor the traffic state 
dynamically. As the development of GPS technology and 
wireless communication technology, vehicles can be equipped 
with GPS devices and their instantaneous positions can be 
transmitted at regular intervals to a central site. When the 
positions of a sufficient number of vehicles can be frequently 
communicated to a central site, travel times can be directly 
measured. This is called Floating Car Data or FCD for short (S. 
Turksma, 2000). FCD have become a very important data 
source for the establishment of a traffic information system. 
The positioning information of FCD sent to the traffic centre is 
geodetic coordinates in the WGS84 coordinate system. 
However, multi spectral remote sensing imagery data is always 
with a projected coordinate system. Therefore, to integrate these 
two together, they must be registered into a same coordinate 
system in advance. There are about 5000 taxis already equipped 
with GPS receivers and wireless communication devices in 
“Shenzhen Urban Transport Simulation System” (SUTSS) 
project. Since May 2006, millions of GPS data sets from these 
5000 taxis have been recorded. This FCD source and high 
spatial resolution image are used to explain the feasibility to 
integrate these two data together. Coordinate systems of these 
two data are listed in table 1. 
In this example (Table 1), the coordinate system of FCD is 
WGS84 geodetic coordinate system, while the remote sensing 
image is projected by simple cylindrical projection method with 
a WGS84 datum. For convenience, FCD can be projected with 
simple cylindrical projection method and then these two data 
are registered in the same spatial reference. The area between 
22°31'48.00"N and 22°32'24.72”N in latitude, 113°54'34.86"E 
and 113°55T3.56"E in longitude is selected as the study area. 
FCD is collected from June 3 rd to June 5 th , 2007. Figure 2(a) 
and Figure 2 (b) respectively shows the images before and after 
the overlay of FCD and high resolution remote sensing image. 
From figure 2, it is obvious that after the coordinate 
transformation, FCD matches the remote sensing image very 
well and FCD almost covers all the roads. By measuring the 
distances from each FCD to its corresponding central line of 
road and do a statistics analysis, it is found that accuracy of 
FCD position is commonly within 10M and a large number of 
FCD has accuracy higher than 4M. The mean deviation of 
arithmetic mean is 2.28M, sample variance is 6.92 and standard 
deviation is 2.63M (Fan, 2007). In the mean while, the spatial 
resolution of this remote sensing image is around 1M. Thus, 
taking into account the width of road, it is convincible that these 
two data can be matched very well. 
Data Type 
FCD 
RS image 
Coordinate System 
WGS84 Geodetic Coordinate 
System 
Simple Cylindrical projection with 
a WGS84 datum 
Tablel Coordinate systems of FCD and RS imagery