Full text: Proceedings, XXth congress (Part 1)

    
  
  
   
    
     
    
   
     
     
    
   
    
     
    
    
   
    
  
   
     
   
  
   
  
   
    
    
   
     
      
   
      
      
     
  
   
   
   
    
     
   
   
    
   
  
   
Istanbul 2004 
e overall rms 
ection. 
age-matching 
A through a 
images. This 
alized cross- 
ding pixels in 
difference in 
, or parallax, 
converted to 
a level datum 
Geomatica is 
stem operates 
osition in the 
m gray level 
| is computed 
cient and the 
iC. correlation 
0 32 pixels at 
at every pixel 
is performed 
vert the pixel 
© matching of 
els where the 
pixel in the 
s. In case of 
terpolate and 
-[DEM ) thus 
1 with respect 
ct the output 
Triangulation 
GN ellipsoid 
rojection has 
imetric rules 
ly, dz, rw, rj, 
'GS84. rw, rj 
WGS84, in 
VGS 84. The 
02] meters. 
JEM surface 
been created 
eine.  Erdas 
chnique with 
hnique also 
Ms that have 
S 
ns have been 
re: 
s (BDTOPO 
erence DEM 
¢ BDTOPO 
ial have been 
  
5.1.1. Comparison of Reference DEM Datasets (Laser DEM 
& BDTOPO DEM): Both the DEMs have been imported into 
native format of Erdas Imagine. The affine transformation 
option of Erdas Imagine has been used to transform the 
BDTOPO DEM at 5m for comparing both the DEMs on same 
pixel resolution. A common area on both the DEMs have been 
extracted using the subset option of Erdas and the two sets of Z 
values compared. 
Table 1: Difference Statistics of Laser DEM and BDTOPO 
DEM for Montmirail 
Range of Height Difference: -20 to 27 m 
  
  
  
  
  
  
  
Difference (in | Percentage Cumulative 
Meters) of Pixels Percentage 
of Pixels 
Up to | m 60.53 60.53 
Up to 3m 31.61 92.14 
Upto5m 05.88 98.02 
Up to 7m 01.42 99.44 
Above7 m 00.56 100.00 
  
  
  
  
As per the specifications of both the DEMs the accuracy is 1 m 
rms. The statistical result shown above indicates that the DEMs 
differ by more than ] m. Only 60.5 percent of the pixels are 
under | m difference. Since the input sources for DEM 
generation are different, there may be some difference in the 
results. The other cause of difference may be due to digital 
surface model (DEM + Canopy) output in case of Laser DEM. : 
5.1.2 Comparison of Saphire DEM (SACDEM) with 
Reference DEMs: The Saphire DEM and the reference DEMs 
have been imported into the native format of Erdas Imagine. 
Since the Saphire DEM has been generated in Geographic 
projection system, it has been re-projected in Lambert-2 NTF 
format with Clarke 1880 IGN ellipsoid. A common area on the 
three DEMs have been extracted and Z values compared. 
Results are tabulated in Tables 2 and 3. 
Table-2: Difference Statistics of Saphire DEM with 
BDTOPO DEM (10 m resolution) after compensating a 
vertical bias of 30 m 
  
  
  
  
  
  
  
  
  
  
  
  
Difference (in Percentage | Cumulative 
Meters) of Pixels Percentage 
of Pixels 
Up to 10m 67.34 67.34 
>10 mand <= 20 m 17.05 84.39 
>20 m and < = 30 m n I 91.50 
230 m and < = 40 m 3.33 94.83 
>40 m and < = 50 m 1.80 96.63 
>50 m and <= 60 m 1.01 97.04 
Above 60 m 2.96 100.00 
  
5.1.3 Comparison of PCIDEM and Reference BDTOPO 
DEM: Both the DEMs have been imported into native format 
of Erdas Imagine. A common area on both the DEMs have been 
extracted using the subset option of Erdas. The PCIDEM was 
generated for the maximum overlap area but has been extracted 
according to the available BDTOPO DEM. Initially the DEM 
was generated in UTM projection with WGS 84 spheroid and 
datum. For comparison the DEM was re-projected into LCC 
Clarke1880 IGN projection with NTF datum using Erdas 
Imagine standard software re-projection tool. The comparison 
results are much poorer (up to 20 m accuracy for 54% of points 
and up to 10 m for only 30% of points) and hence not included. 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004 
457 
Table-3: Difference Statistics of Saphire DEM with Laser 
DEM (5 m resolution) 
  
  
  
  
  
  
  
  
  
  
  
Height Percentage Cumulative 
Difference of Pixels Percentage 
(in Meters) 
Up to 1 m 24.26 24.26 
Up to 2 m 15.77 40.03 
Upto3m 6.02 46.96 
Up to 4 m 6.11 53.07 
Up to 5 m 5.20 58.26 
Up to l0 m 17.11 75.37 
Up to 20m 13.80 89.17 
Up to 30m 5.10 94.27 
Above 30m 373 100.00 
  
  
  
  
Figure-1 gives a comparison of profiles from Saphire DEM 
(SACDEM), PCIDEM and BDTOPO DEM in 10 m resolution. 
Figure-2 gives a comparison of profiles from Saphire DEM 
(SACDEM ) and LaserDEM in 5 m resolution. 
Figure-1: A Diagonal Profile of BDTOPO DEM, Saphire 
DEM (SACDEM) and PCIDEM (10 m resolution) 
Comparison Graph Without Compensating Bias 
  
  
  
   
  
300 
450 
400 
350 
E30 — TOPODEM 
5 250 -——— —SACOEN | 
$ 200 RD | 
I 
130 
100 
30 
0 
0 1000 2000 3000 4000 5000 6000 7000 8000 
Distance (m) 
Figure-2: A Diagonal Profile of LaserDEM and Saphire 
DEM (5 m resolution) 
Comparison Graph (LaserDEM Vs SaphireDEM) 
600 
  
  
500 Pry 
450 X 
L 
7 4 
; &. A EN. ur am 
en f / Ä 77 v uti SEN TI 
: y. N N 
d —LaserDEM —— 
Ar = 
Saphire DEM. 
4 ..Süpilre UEM. 
  
  
  
  
  
  
0 1000 2000 3000 4000 3000 6000 7000 8000 3000 10000 
Distance (m) 
5.2 Melbourne Data Set: The DEM for this data set could not 
be generated using PCI Geomatica software possibly due to the
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.