The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
247 
Natural river profile statistics 
Figure 1. Natural river’s profiles sampled on different sections 
Artificai river profile statistics 
Figure 2. Artificial river’s profile sampled on different sections 
(the sampling points with a certain distance) 
Sampling points 
Figure 3. Highway’s profile sampled on different sections 
(the sampling points with a certain distance) 
sophisticated computation. This paper suggests using the logical 
relationship of the locations for different height sections on the 
profile, which can simplify the judgement process. 
The following consideration about establishing the concrete 
judgement functions takes the skeleton as the initial reference, 
and ranges 2 times the distance from central line to the 
extracted edge. Based on the sum up about the typical forms of 
the profile distribution in the upper three figures, the logical 
relationship between initial position, end position, highest 
position and lowest position can be determined and be used for 
making sure of different profiles. Then the discrimination 
function can be summed up. 
The profile factor logical function of natural river, highway and 
artificial river can be summed up as formula (5), (6) and (7) 
individually. If the real profile fits one of the conditions, then 
the object can be determined as the related terrain catogery. 
The profile morphological factors can be expressed as shape 
functions forjudging. Highways are commonly with gutters and 
the road surface is higher than the fields both sides, so its 
profile can be united into “W” shape. Natural rivers are like 
“U” without banks, while artificial water-bodys have riverbanks 
and its profile can be simplified as “M” figure. Then 
corresponding profile factor functions can be established for 
calculation and judgment accordingly. 
P = P 
max lnit 
P = P 
max End 
P < P < P 
1 lnit min 1 End 
(5) 
(6) 
2.3 Profile functions 
For automatic identification, the distribution of rivers’ profiles 
should be summarized into functions for logical judgement. 
Natural rivers, artificial rivers and highways are individually 
deduced as the shapes shown in figure 4. For highway and 
artificial river, the upper shapes are more common, but the 
lower ones still exist. 
The concrete mathematical expression can assume sine function, 
which distinguishes by the fitting range in different phase and 
amplitude. The shape deployment also can be characterized by 
different degree polynomial functions with various coefficients 
as detection parameters. But these kinds of method above need 
P = P 
min lnit 
P = P 
mm 1 End 
P <P <P 
1 lnit 1 max 1 End 
(7) 
where P init = height of the point on the centreline 
Pend = height of the point at the end of the profile 
Pmax = height of the point with largest elevation 
P min = height of the point with least elevation 
The above discrimination functions need some pre-processes to 
ensure the reliability of the practical judgement. Firstly, the data 
for implementation should assume the average result of