tributed in the whole country, using computer net- 
work techniques organize the railway transportion 
work in facilitating the transfor of information from 
one area to another. It is dynamic monitoring the 
whole national moved train cars in real—time. This 
optimum transportation path selected system is one 
program of the RTMIS. It can choose the shortest 
and lowest cost transportation path. A customer 
may ask the following queries: 
1) How find the shortest transportation path from a 
given two railway station for saving the transporta- 
tion time to keep fresh of the cargo (such as 
vagetable, fruits...) 
2) Find the nearest railway station from a given fac- 
tory for convenience to transport the factory’ cargo 
to the nearest railway station. 
3) Find the lowest transportation cost path to save 
the money. 
The transportation management units may ask the 
following queries: 
4) Show the region in which the nearest railway 
station is with 100 kilometer. This area call the at- 
tractive area of cargo and passengers. In different 
radius can inventory all important transportation in- 
formation data such as big factories, business, large 
cities and villages. .. etc. The railway transportation 
planner can estimate the volume of the cargo and 
passengers. 
5) Show the region in which the nearest railway 
station is the railway station at S;. 
6) Find the nearest rolling stock plant from a give 
railway station. If some train cars break down, it 
can send to the nearest rolling stock plant to repair. 
7) If the shortest transportation path some section 
of railway line interrupt traffic by nature disaster, 
find the second shortest transportation path or kth 
shortest transportation path. 
These queries Q2, Q4, Q5, and Q6, can be an- 
swered by the Voronoi diagram (Voronoi 1908). 
For example, in Figure 2, given a set of railway 
S,. We define a 
buffer zone, the region enclosed by the circular 
Let 8,, 8,, ... S, be a finite number of dis- 
tinct points in the 2— dimensional Cartesion Space, 
F be arbitrary factory location, d (F, S;) be the 
Euclidean distance between location F and S;. If 
stations located at S,, S,, 
arcs. 
the set of location satisfying the condition that 
d(F, 8) <d(F, 8) 
980 
for all j except for ? — j. The 5, is the nearest 
railway station from F . Queries Q2, Q4, Q5 are 
directly given by the Voronoi diagram. For exam- 
ple, in Figure 2, the region in which the nearest 
railway station is the one at S, is given by the 
shaded region. 
To answer Q1 and Q3, we consider a national rail- 
way transportation network N. Consisting of con- 
nected line segment (called links), and a set of rail- 
way stations. 
S zm 4S. So. B.) 
on the define the 
distance d,4, (S, S;) froma railway station S to S; 
by the length of shortest path from S to S; called 
the railway network distance. The Voronoi diagram 
on a railway network which gives the answer to 
these queries. The Voronoi diagram is defined by 
the set V,,CS;) of location on N , from railway sta- 
railway network N, we 
tion 5 the network distance to railway station S; is 
less than or equal to the network distance to any 
other points i.e. for S; € N . 
Via (8) = (8, 8, € N, d,,(8, 8) <d(8, 8) 
j263, = |, 2, 3... (1) 
The first queries is easily answered if the railway 
rR 
network Voronoi diagram is obtained. The compu- 
tational method for constructing this Voronoi dia- 
gram is fairly straightward if we can use a shortest 
path algorithm (Ford and Fulkerson 1962). 
In different railway line, different cargo, the trans- 
portation price are different. The unit transporta- 
tion price is defined price of one ton cargo transport- 
ed one kilometer, {; , we can modified (1) as fol- 
lowing 
V aa (83) = (5i; 5; EN, 
tid (8S, 8) td CS, 8) (2) 
The lowest transportation price on a given two rail- 
way stations are obtained by weighted Voronoi dia- 
gram. 
4. Conclusions 
In this study, the automatic railway design system 
is work. The results of selected railway line by com- 
puter is near the results by traditional method. But 
this system is more efficiency and saving a lot of 
time. More experiments will be done to meet the 
different terrain model. 
The second spatial analysis model studying work 
just begin. In practice, basic work is to establish a 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996