The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
X'^X' + 'ZXrÜX-X'^a-T, 
j=1 
(9> 
7=1 
= ^0+'y'.z j- tj ,j Z i — Z ( . | < y • T z 
7=1 
In equation (9), (X’^Y’^Z’,) is the predicted position of 
(Xj,Yi,Zi), and (T x ,T y ,T z ) are the given 3D spatial thresholds, 
(a,p,y) are coefficients that calculated according to the steps 
from the initial place. These linear varying coefficients are 
calculated by the check points in the field work route. 
3. DATA 
The mobile mapping field work is done on Nov 2007 in 
Shanghai, the biggest city of China. The experiment data 
collected from the period of Yan’an Road. It covers the typical 
surroundings of metropolis, such as overhead road, viaduct 
crossing, large amount of skyscrapers along the two sides of 
road, heavy and variable transportation and so on. These factors 
do complicated influence to the GPS signals and wireless 
communication signals of network RTK together. There is 1858 
network RTK position points is obtained altogether along the 
road. From the concrete check, these points include different 
kinds of solutions, such as pseudo-range GPS, differential GPS, 
RTK and so on. The smallest time interval between two data is 
1 second. The whole contribution is expressed as the following 
figure (see Fig.l). 
Hongkou 
District - 
Zhab» District 
Jing'an District 
Putuo District 
Nan shi 
. District 
lu wan 
District 
Changning District . 
Xu hut District 
Mmgtiang District 
* blue curve expresses the whole route of network in (a); (b) is the red rectangle zoomed in (a) and blue points in (b) is the network RTK data. 
Fig.l Mobile mapping’s network RTK data on Yan’an Road 
4. EXPERIMENT AND RESULT ANALYSIS 
As we know, network RTK position data’s accuracy and 
reliability are affected by GPS signal and wireless 
communication network, and mobile mapping platform is 
always in moving state. Then the actual network RTK data 
always include differential GPS with different base of the base- 
network, and single point positioning result. Besides, GPS is 
always much better on the horizontal plane positioning than the 
vertical plane. In the filtering experiment, horizontal speed and 
vertical speed are considered separately. The following figure 
(see Fig.2) shows the result comparing to the original data (red 
points show the filtered result, blue points is the original which 
means filtered). 
Fig.2 shows a part of filtered result of network RTK points. 
From the result, it is easy to know that the filtering work can 
smooth the error of network RTK. There are 1858 network 
RTK points and the remains after filtering processing are 1467, 
1532 and 1587 points with three different thresholds (see Tab.l). 
In order to simplify the filtering difference and find the 
common knowledge in different kind of network RTK data in 
mobile mapping, the variable thresholds of XYZ are set in the 
same scale each time. There are five kinds of point solutions in 
network RTK mode: solid unavailable, pseudo-range, 
differential GPS, solid RTK and float RTK. In general 
surveying work, the accuracy is from low to high as the above 
order. From Tab.l, we find that positioning data filtered are not 
only marked as pseudo-range point positioning and float 
differential GPS in network RTK data, but a rather big partition 
is the class of solid differential positioning result (It is shown 
more clearly in Tab.l); what is more, each kind of network 
RTK point maintains a quite steady portion in different given 
filtering threshold. So it means that the prior knowledge about 
accuracy of pseudo-range positioning is low, differential GPS 
and RTK positioning is high is not absolutely reliable in mobile 
mapping work in dynamic measurement as mobile mapping, 
and more research should be done in the complicated 
relationship in GPS dynamic measurement and communication 
factors in mobile mapping work. 
Fig.2 Filtered result of network RTK based on platform 
dynamic characteristic