Case Study on Point Data - Open Source Geospatial Tools: Applications in Earth Observation - page 315

We start by applying the first algorithm that is based on a progressive morphology
filter (Zhang et al. 2003) and implemented in spdpmfgrd :

spdpmfgrd -r 50 --overlap 10 --initelev 0.1 --maxfilter 14 --grd
→ 1 -b 0.5 -i LiDAR_10m.spd -o LiDAR_10m_pmf.spd

-r 50
Number of rows within a block to is 50.
--overlap 10
Size (in bins) of the overlap between processing blocks is 10.
--initlev 0.1
Initial elevation difference threshold.
--maxfilter 14
Maximum size of the filter is 14.
--grd 1
Threshold for deviation from identified ground surface for classifying the ground
returns is 1.
-b 0.5
Bin size for processing and output image is 0

. 5m.

-i LiDAR_10m.spd
Name of the input SPD point dataset.
-o LiDAR_10m_pmf.spd
Name of the classified SPD point dataset.

The output is then used as input to a multiscale curvature classification (Evans
and Hudak 2007), implemented in spdmccgrd .

spdmccgrd -r 50 --overlap 10 --class 3 -i LiDAR_10m_pmf.spd -o
→

LiDAR_10m_pmfmccgrd.spd

-r 50
Number of rows within a block is 50.
--overlap 10
Size (in bins) of the overlap between processing blocks is 10.
--class 3
Only use points of class 3.
-i LiDAR_10m_pmf.spd
Name of the input SPD point dataset.
-o LiDAR_10m_mfmccgrd.spd
Name of the output.