High resolution digital terrain models (DTMs) created from interpolation of 3D point clouds from airborne laser scanners can suffer from "noise" attributable to different causes. Noise is usually defined as a random error (white noise), thus with no bias (zero mean) and finite standard deviation. In this presentation we assume a broader sense for noise, including any other causes whose effect is the deviation from a correct representation of the ground surface. It is known that DTMs which are derived from filtered and interpolated laser scanning data have low geo-morphological quality (Kraus and Pfeifer, 2001), with spurious pits in valleys which decrease the quality of hydrological analysis. LiDAR surveys with high pulse frequencies (used to obtain high point densities) sometimes do not have an homogenous distribution of the point density, thus resulting in lower quality interpolation. In the following presentation a comparison is carried out of methods which increase the geo-morphological quality of DTMs in mountainous areas. Results show that post-processing with a despeckling method which uses the best orientation of normals (Sun et al., 2007; Stevenson et al., 2010) has the significant characteristic of preserving significant features while virtually eliminating most of the noise. The study compares also the effect of threshold over the method, concluding that a threshold of 0.4 gives best results for obtaining high-resolution DTMs with high geo-morphological quality in mountainous areas. This study can be helpful to document how to process DTMs which have a lower quality due to noise factors.

Denoising Methods to Improve Lidar-Derived High Resolution Digital Terrain Models.

PIROTTI, FRANCESCO;GUARNIERI, ALBERTO;VETTORE, ANTONIO
2011

Abstract

High resolution digital terrain models (DTMs) created from interpolation of 3D point clouds from airborne laser scanners can suffer from "noise" attributable to different causes. Noise is usually defined as a random error (white noise), thus with no bias (zero mean) and finite standard deviation. In this presentation we assume a broader sense for noise, including any other causes whose effect is the deviation from a correct representation of the ground surface. It is known that DTMs which are derived from filtered and interpolated laser scanning data have low geo-morphological quality (Kraus and Pfeifer, 2001), with spurious pits in valleys which decrease the quality of hydrological analysis. LiDAR surveys with high pulse frequencies (used to obtain high point densities) sometimes do not have an homogenous distribution of the point density, thus resulting in lower quality interpolation. In the following presentation a comparison is carried out of methods which increase the geo-morphological quality of DTMs in mountainous areas. Results show that post-processing with a despeckling method which uses the best orientation of normals (Sun et al., 2007; Stevenson et al., 2010) has the significant characteristic of preserving significant features while virtually eliminating most of the noise. The study compares also the effect of threshold over the method, concluding that a threshold of 0.4 gives best results for obtaining high-resolution DTMs with high geo-morphological quality in mountainous areas. This study can be helpful to document how to process DTMs which have a lower quality due to noise factors.
2011
Epitome
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11577/153538
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact