Airborne laser scanner is widely adopted for city modeling, DTM (Digital Terrain Model) generation, monitoring
electrical power lines and detection of forest areas. In generally, airborne laser scanning enables to acquire point cloud
3D data for surface of the ground or objects using multiple return pulses (first, last and other pulse). Filtering for
distinguish on- and off-terrain points from point cloud 3D data which are collected by airborne laser scanner was issue,
and various filtering methods have been developing for generating DTM using point cloud 3D data.
Waveform information (range, pulse amplitude, pulse width) which is corrected by resent laser scanner system has been
receiving more attention for improvement of classifying the point cloud data into on- and off-terrain points. Waveform
information has ability to classify the point cloud data, however, robust filtering for distinguish on- and off-terrain points
is still issue. The main problem is robust extraction of the deepest points which shows ground surface. As many filtering
and classifying methods for robust extraction of the deepest points were proposed including waveform information, the
problem reaches extraction of the last pulse since the last pulse show the deepest points.
With this motive, filtering and classifying approach for DTM generation in forested area using multiple return pulses
instead of waveform information are investigated in this paper.