vor etwa 14 Monaten habe ich die Programmierarbeiten an einem Projekt abgeschlossen und bis gestern war ich damit beschäftigt das Ganze auf Papier zu bringen. Natürlich darf ich Euch diesen Schmachtfetzen edelster Literatur nicht vorenthalten. Daher, exklusiv, aus der Einleitung!
Terrestrial 3D laser scanning (TLS) has been utilised in various applications where geometric information needs to be captured with high spatial density such as as built documentation, archaeology, cultural heritage, mining and reverse engineering (Boehler et al., 2004, Bae et al., 2007, Gordon, 2005). 3D laser scanners are capable of capturing geometric information, colour information (RGB) for objects and also the strength of the returned laser beam. Among these information from TLSs, the usefulness of the intensity value is mainly for target detection as well as for visualising a measured point cloud (Lichti, 2002). However, raw intensity return of laser scanners is needed to be calibrated (Pfeifer et al., 2007)
.
The scientific focus in TLSs has been mainly set onto the determination and investigation of the accuracy as well as the extraction of geometrical features from point clouds (Belton, 2008). Few research projects have investigated laser scanners’ intensity value. For example, Hoefle and Pfeifer (2007) utilised a model-driven approach to correct the influences of topographic and atmospheric effects on the intensity value of airborne laser scanners (ALS). In addition, Kaasalainen et al. (2007) developed a rigorous approach to classify the surface of glaciers by using the point cloud information including calibrated ALS intensity values. Lichti (2002) analysed the effects of reflecting surface material and investigated the relationship between intensity and distance in order to classify 3D point clouds with the correctly calibrated intensity. Hancok et al (1998) used terrestrial laser scanners’ intensity for the detection of obstacles for highway environments to control mobile robots. Lichti (2005) used near infrared intensity in combination with colour information to filter and classify terrestrial laser scanner point clouds. Wang et. al. (2009) described the potential of intensity data for planar surface segmentation.
The scientific focus in TLSs has been mainly set onto the determination and investigation of the accuracy as well as the extraction of geometrical features from point clouds (Belton, 2008). Few research projects have investigated laser scanners’ intensity value. For example, Hoefle and Pfeifer (2007) utilised a model-driven approach to correct the influences of topographic and atmospheric effects on the intensity value of airborne laser scanners (ALS). In addition, Kaasalainen et al. (2007) developed a rigorous approach to classify the surface of glaciers by using the point cloud information including calibrated ALS intensity values. Lichti (2002) analysed the effects of reflecting surface material and investigated the relationship between intensity and distance in order to classify 3D point clouds with the correctly calibrated intensity. Hancok et al (1998) used terrestrial laser scanners’ intensity for the detection of obstacles for highway environments to control mobile robots. Lichti (2005) used near infrared intensity in combination with colour information to filter and classify terrestrial laser scanner point clouds. Wang et. al. (2009) described the potential of intensity data for planar surface segmentation.
This paper presents a calibration method for the intensity values from 3D laser scanners. Firstly, four experiments with the laser scanner (Leica HDS 3000), a camera (Nikon D2Xs) and a monochromator (LOT Oriel Lambda 500) were conducted in order to investigate the relationships between the returned laser beam and the incidence angle of the laser, the distance between a laser scanner and objects as well as the radiometric properties of the objects. From these instruments, a calibration model for the laser scanner (Leica HDS 3000) was presented. Second, in order to demonstrate the effectiveness of this approach, a segmentation method was used for an uncorrected and a corrected point cloud.
Keine Kommentare:
Kommentar veröffentlichen