Experimental comparison of filter algorithms for bare-Earth extraction from airborne laser scanning point clouds

https://doi.org/10.1016/j.isprsjprs.2004.05.004Get rights and content

Abstract

Over the past years, several filters have been developed to extract bare-Earth points from point clouds. ISPRS Working Group III/3 conducted a test to determine the performance of these filters and the influence of point density thereon, and to identify directions for future research. Twelve selected datasets have been processed by eight participants. In this paper, the test results are presented. The paper describes the characteristics of the provided datasets and the used filter approaches. The filter performance is analysed both qualitatively and quantitatively. All filters perform well in smooth rural landscapes, but all produce errors in complex urban areas and rough terrain with vegetation. In general, filters that estimate local surfaces are found to perform best. The influence of point density could not well be determined in this experiment. Future research should be directed towards the usage of additional data sources, segment-based classification, and self-diagnosis of filter algorithms.

Introduction

While commercial airborne laser scanning systems have come a long way, processing the huge point clouds for the purpose of modelling systematic errors, filtering, feature extraction and thinning often requires a large amount of human interaction. For the production of digital elevation models, the manual classification (filtering) and quality control pose the greatest challenges, consuming an estimated 60–80% of processing time (Flood, 2001), and thus underlining the necessity for research in this area. Algorithms have been developed for semi-automatically/automatically extracting the bare-Earth from point clouds obtained by airborne laser scanning and InSAR. Some comparison of known filtering algorithms and difficulties have been mentioned in Huising and Gomes Pereira (1998), Haugerud and Harding (2001) and Tao and Hu (2001). However, an experimental comparison was not available, although it would be useful to assess the strengths and weaknesses of the different approaches based on available reference data. Therefore, the ISPRS Working Group III/3 “3D Reconstruction from Airborne Laser Scanner and InSAR Data” initiated a study to compare the performance of various automatic filters developed to date, with the aim of:

  • determining the comparative performance of existing filters,

  • determining the influence of point density on the filter performance, and

  • identifying directions for future research on filtering point clouds.

In line with these aims, a web site was set up at which datasets were provided for testing. Individuals and groups wishing to participate in the study were kindly requested to process all datasets, if possible. Results were received from eight participants. The algorithms used by the participants come from a cross-section of the most common strategies (or variants) for extracting the bare-Earth from airborne laser scanner point clouds and thus provide a good overview on the state-of-the-art.

This paper is structured as follows: The laser scanning data used in the test is described in Section 2. 3 Evaluated filter algorithms, 4 Filter characteristics describe the evaluated filter algorithms and try to classify their characteristics. The results of the experiments are described and discussed in Section 5. In Section 6, the conclusions are drawn with respect to the objectives set out.

Section snippets

Test data

Within the framework of the OEEPE project on laser scanning (OEEPE, 2000), FOTONOR acquired data with an Optech ALTM scanner over the Vaihingen/Enz test field and the Stuttgart city centre. With kind permission of the OEEPE, subsets of this dataset were selected for the comparison of filtering algorithms. Reference data was produced by interactively filtering the datasets.

Evaluated filter algorithms

Eight individuals/groups submitted results for the test. An overview of the participants and a characterisation of their filters are given in Table 1. The algorithms are described in more detail below.

Filter characteristics

Filters are built from combinations of different elements. From a study of the submitted algorithms and others published in papers, seven elements were identified.

Results and analysis

The filter results of the participants have been analysed in various ways. The data of all eight test sites has been used to visually assess the performance of the algorithms in several difficult terrain types (Section 5.1). This more qualitative analysis was followed by a quantitative analysis using the 15 subsamples that dealt with specific cases (Section 5.2). Furthermore, the effect of the point density on the performance of the filter algorithms was assessed quantitatively (Section 5.3).

Discussion

The objectives of the study were to: (1) determine the performance of filter algorithms, (2) determine how filtering is affected by point density and (3) establish directions for future research. These objectives are treated individually in the sections below.

Acknowledgements

This study would not have been possible without the help and cooperation of participants who took time from their schedules to filter the 12 datasets. The authors wish to extend their gratitude to Peter Axelsson, Christian Briese, Maria Brovelli, Magnus Elmqvist, Norbert Pfeifer, Marco Roggero, Gunho Sohn, Roland Wack and Andreas Wimmer.

References (26)

  • M. Flood

    LIDAR activities and research priorities in the commercial sector

    International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

    (2001)
  • R.A. Haugerud et al.

    Some algorithms for virtual deforestation (VDF) of LIDAR topographic survey data

    International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

    (2001)
  • K. Kraus et al.

    Advanced DTM generation from LIDAR data

    International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

    (2001)
  • Cited by (889)

    • G<sup>1</sup> spline functions for point cloud fitting

      2024, Applied Mathematics and Computation
    • Towards intelligent ground filtering of large-scale topographic point clouds: A comprehensive survey

      2023, International Journal of Applied Earth Observation and Geoinformation
    View all citing articles on Scopus

    This paper is an extended and improved version of the paper presented at the ISPRS WG III/3 workshop on “3D Reconstruction from Airborne Laser Scanner and InSAR Data”, Dresden, Germany, 8–10 October 2003.

    View full text