The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
428 
The legend for the figure, purple is the lowest height. 
Figure 3. A (right) data gap, (left) road with a bridge and trees, B pool, C and D large and irregularly shaped buildings, E a crossroad 
with small tunnel, F bridge. (Illustrated by Cloud Peak Software LASEdit Utility Copiright 2006 All Rights Reserved) 
The test site has a crossroad with a tunnel, a bridge, a pool, 
irregular shaped buildings and trees.The planimetric 
resolution is 0,67 points per square; thus, the two points 
spacing is 1-1.5m (Figure 3). 
2. M ETHODS 
The methods presented here for tree extraction are related to 
mathematical algorithms: 
1. Preprocessing of the raw laser data (true DSM) 
2. Application of mathematical algorithms 
3. Building of images. 
The input file “cite.txt” has 8 collumns and 243.400 rows. 
The collumns are X, Y and Z coordinates (Easting, Northing 
and Height) and intensity of first returns and X, Y and Z 
coordinates (Easting, Northing and Height) and intensity of 
last returns in that form: 
XI Y1 Z1 II X2 Y2 Z2 12 
513450.03 5402650.22 296.38 3 513450.04 5402650.23 296.30 3 
513450.04 5402651.74 295.44 146 513450.04 5402651.74 295.44 146 
513449.85 5402653.04 301.16 0 513450.00 5402653.29 296.12 80 
Figure 4. Due to diffusion in trees only one or two of the rays 
come back to detector, resulting intensity drop. 
Intensity Drop Method: When the laser light hits trees, there 
is loss in intensity diffusion(Figure 4). The mathematical 
algorithm is based on intensity drop. 
The condition to find trees is the three following intensities 
be less than treshold value. After tests the treshold value was 
chosen 35 (Figure 5).