Istanbul 2004 International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
  
  
  
  
  
  
the digitized the largest non-terrain object. Finally, the coarse DTM is integrating information about terrain relief, slope and data 
tation height refined hierarchically from the top level to the bottom level. At density. The non-terrain points are replaced by interpolated 
e remaining each level, denser terrain points are identified, and a smooth elevations using surrounding terrain points. The bottom-level 
condition is employed to improve the processing efficiency by image represents the expected bare Earth surface. 
MA Airfield [ Digital Elevation Data: Lidar | | Digital Photo Data: Aerallmage | GIS Data: Topographical IM. 
that protrude S - | | ry paw eT | 
| Specification: Airfield Init. Doc. | 
Vehicular 
OIS Y 
  
Y 
( Scanning | 
  
| T 
e regard as Filtering Filtered Qutliers 
ions, one of Ÿ 
g, 2002), is F- Lidar DSM 1 Comparson& | 
  
   
  
  
        
cation levels, Digitization 
vel must be 
  
  
  
Digital Image Processing 
  
  
  
     
  
  
   
  
  
  
  
  
  
   
  
  
  
   
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
  
  
  
Traverse = G ic Resistration & 
à ; r Surfaces Geometnc Registration & 
the National seen Ways | Correction using GCPs 
buildings, or DEM/DTM 1 Y = : 
ns and need ( Manual Digitization m S Bilinear Interpolation 
his level are ( Elevation Subtraction ] [ RC] Comparison & ECL 3 
t uildinsss Sr Vist] Airfield Backgroun 
OIS-related EET Buildings shp i Digitization | Airfield Background — | 
runway and ideni : 
toi vi ects Y. | | Residential Areas shp | | Y Spatial Selection & 
C s 8018, Calculation asie iL. 7 d Data Conversion 
f factors are : [ Bissein Moret | BARS: 
ifvi Trees/Bush oads shp Ÿ 
ifying and Datasets Grids Vector/Raster Conversion Rivers shp 
I 
v Y Four Raster Layets 
[- Subtraction f Raster Calculation ] 3 Distance 
| Runway Center Lines | icai 
Em 
rate Buffered C. i Sm 
Hiusteaies [ RasterfVector Conversion ] ( bie Corer Lines J Penetration 
lidar data ^w = Ÿ 
plies image , a Saaty Risk Modeling: 
le the DTM '' - aty Risk Modeling: 
e ( Datasets Ivlerger ] |. 3D Visualization of 2 Gustuction Raster Reclassification 
    
  
ildings and 
he forested Obstructions shy Add Attributes 
lential areas 
IS surfaces, 
Map Calculation 
| Airfield Obstruc tions “Identification | ^ ; 
"nn í,, "m T Sequencing & Rendering 
  
  
  
  
  
SVTW. The Figure 2. Project workflow 
nd bilinear 
new map is In Figure 3, TIN is used to visualize the continuous surfaces of ; 
erge all the the DSM (3a) and DTM (3b) (Burrough and McDonnell, 1998). DSM-Diagonal 
be recorded 
'rs make a P 150 
s are added. d 
the finally s — 100 
ulation and |, m 
| qd 
| = #5 
| ü 
:ct products Ü 5000 10000 —— 15000 
oint clouds Ricco a 
covered by NI or lip oA Profile Graph 
ig, the raw 
or blunders, 
' very large 
A median DEM-Diagonal 
r removing | E pT QE 150 
e by which | joe hal 
the central u 
£. 100 
Figure 3b. 3D Visualization of the DTM gom 
ical terrain , = 50 
algorithm Figure 4 shows the two profiles along with the pink diagonal 
and other lines in Figures 3a and 3b, respectively. It can be observed that ff 
ot 
  
  
rface in a the DSM surfaces are more complex than the DTM's. This is d CODD 10000 Tan 
unders, the because the DSM includes all non-terrain objects, but the DTM 
ased range shows the bare terrain surface only. Profile Graph 
| each grid. : 
el image is When the DTM is generated, we also obtain an estimate to the dea 
larger than DNM by subtracting the DTM from the raw lidar range image. Figure 4. Profiles in the DSM (top) and the DTM (bottom)