An Adaptive Filtering and Terrain Recovery Approach for Airborne LiDAR Data

International Journal of Innovative

Computing, Information and Control ICIC International c_{°2008 ISSN 1349-4198}

Volume 4, Number 7, July 2008 pp. 1783—1796

AN ADAPTIVE FILTERING AND TERRAIN RECOVERY APPROACH

FOR AIRBORNE LIDAR DATA

Lianghwei Lee

_{, Mingjer Huang}

_{, Shiahnwern Shyue}

_{and Chengyi Lin}

National Kao-Hsiung University of Applied Sciences Kaohsiung 80778, Taiwan

lhlee@cc.mail.kuas.edu.tw

2_{Department of Marine Science}

Naval Academy Kaohsiung 813, Taiwan

(Corresponding author) mjhuang@cna.edu.tw

3_{Department of Marine Environment and Engineering}

National Sun Yat-Sen University Kaohsiung 804, Taiwan swshyue@mail.nsysu.edu.tw

4_{Chung Hsing Surveying Co., LTD}

Tai-Chung, 403, Taiwan cyl@chsurvey.com.tw

Received April 2007; revised September 2007

Abstract. Airborne light detection and ranging (LiDAR) is a new remote-sensing tech-nique that scans the ground surface to obtain high-resolution three-dimensional data. This paper presents a novel, adaptive LiDAR data-filtering algorithm that effectively fil-ters out ground objects and generates a digital elevation model (DEM). Filtering and performance evaluation results indicate that the proposed algorithm not only filters out ground objects in urban, forest, and mixed land-cover areas, but also solves the difficul-ties that are typically encountered when dealing with complex features using surface-based LiDAR data-filtering algorithms.

Keywords: Light detection and ranging (LiDAR), LiDAR data filtering, Digital ele-vation model (DEM), Hierarchical patchwise second-order polynomial surface filtering algorithm (HPSOFA)

1. Introduction. Digital elevation models (DEMs) are digital representations of terrain

height, whereas digital surface models (DSMs) represent the height of objects on the

earth’s surface. These two types of model have wide-ranging applications, including those

related to civil engineering design, disaster prevention, terrain simulation, cyber-city

con-struction, and forest management [7]. In recent years, the main method of automatically

deriving DEMs has involved digital photogrammetric techniques of matching points on

stereo-pair images derived from satellite or aerial observations. However, this method

may also produce inaccuracies in DSMs and DEMs of densely forested areas and heavily

built-up urban areas, in which image matching can fail because of the homogeneous image

color of forests or occlusion problems in urban areas [20].

In contrast, airborne light detection and ranging (LiDAR) is an operationally mature

remote-sensing technique that integrates a laser scanner, an inertial measurement unit

(IMU), and a global positioning system (GPS) into a single instrument mounted on an

aircraft. This recently developed technique allows for the rapid acquisition of a dense