Generating Digital Terrain Model: Joint Use of Airborne Lidar Data and Optical Images. Génération de Modèles Numériques de Terrain Par Fusion de Données Lidar et Image

Generating Digital Terrain Model: Joint Use of Airborne Lidar Data and Optical Images

Génération de Modèles Numériques de Terrain Par Fusion de Données Lidar et Image

Frédéric Bretar Nesrine Chehata 

Institut Géographique National, 2-4 Av. Pasteur 94165 St. Mandé cedex, France

Institut EGID – Université Bordeaux 3, 1 Allée Daguin 33607 Pessac

Page: 
145-159
|
Received: 
30 October 2007
| |
Accepted: 
N/A
| | Citation

OPEN ACCESS

Abstract: 

The knowledge of an accurate topography is a prerequest for monitoring natural hazards and for environmental management (e.g. hydrologic and erosion models). Remote sensing lidar systems are active sensors which record altimeter data of the Earth’s surface as 3D point clouds : from an emitted laser pulse, the receptor detects backscaterred photons from the reflecting target. The altimeter accuracy is often higher than in the case of using stereoscopicconventional techniques, even if the ground density of the data is often lower. Additionnaly, beyond the altimeter information, these data contain spectral information related to the target reflectance in the infra-red domain. Depending on the target geometry, lidar systems can acquire several echos for a single laser pulse. This property is particularly interesting in forests or in areas of urban vegetation since it provides not only the canopy height, but also, under certain conditions, the terrain height under the vegetation layer.

Many algorithms have been implemented to sort out the automatic classification problem of 3D point clouds providing ground and off-ground points as well as the Digital Terrain Model (DTM). This is mainly due to the various aspects of landscapes within a global survey which can include urban, forest or mountainous areas.

This paper presents the generation of DTMs based on the joint use of optical images and lidar data. The study is focused on rural areas where a large scale mapping is a major issue. We propose an algorithm based on a predictive Kalman filter for which the temporal component is replaced by a spacial indexation. The algorithm consists in analyzing the altimeter distribution of the point cloud of a local area in the local slope frame. We insist on the role of local slopes for determining the ground height especially in case of steep slope terrain. We assume that points of the first altimeter mode (lowest points) belong to the terrain. The mean height of these points correspond to the measured DTM value. The final DTM value at a specific position is obtained by a linear combination between the measured value and the predicted value. The predicted value depends on the neighboring pixels through their respective uncertainties. The slopes are also integrated in the predictive filter. The local slope is estimated by robust M-estimator, based on the local first altimeter mode. The final slope for each site depends on the corresponding neighborhood uncertainties. This process makes the slope estmation more robust. Finally, the predictive Kalman framework provides not only a robust terrain surface, but also an uncertainty for each DTM pixel as well as a map of normal vectors.

The local window size is a critical parameter in filtering algorithms. It should be small enough to keep all ground details but large enough to ensure the removal of off-ground objects such as trees or/and buildings. Consequently, we introduce an adaptation of the local neighborhood based on the integration of image and lidar data within a predictor of high vegetation areas. In this context, we decided to investigate the potential of using raw uncalibrated lidar intensity in case of a joint indice computation, which is generally derived from image-based infrared data. Consequently, lidar intensity and optical images are combined to generate a Hybrid Normalized Difference Vegetation Index (HNDVI). The lidar provides the infrared band whereas the aerial image provides RGB spectral bands.

A vegetation mask is then calculated with HNDVI and lidar variance information. In fact, higher values of HNDVI are likely to correspond to vegetated areas and vegetation is described as non-ordered point cloud with a high variance compared to human-made structures. Based on the vegetation mask, the local neighborhood size continuously varies from a predefined minimum distance to an automatically processed upper boundary depending on the vegetated area density. Lidar data have been collected in 2004 by the Institut Français de Recherche pour l’Exploitation de la Mer (IFREMER) over the Morbihan Gulf, France. It has been funded by TOTAL. The entire survey is composed of 230.106 points with intensities and has been acquired with an ALTM (Optech) system 1210. The point density is 0.7 pt/m2. The lidar wavelength is 1064 nm. Optical images are extracted from the BDOrtho ®French orthophoto data basis of the Institut Géographique National with a nominal resolution of 0.5m, but resampled at 2.5 m for the generation of the Hybrid-NDVI image.

Conclusive results show the potentiality of a full combination of lidar data and RGB optical images for improving the generation of fine DTMs on rural environments. The filtering process is more robust, DTM are less picked due to vegetation filtering. However, in case of dense vegetation, the window size can be large and the average terrain height may be underestimated. As future work, further area predictors can be developed to adapt the proposed methodology to various landscapes and relieves.

Résumé

gestion de l’environnement et des risques naturels. En complément de l’image traditionnelle riche en contenu sémantique, la télédétection active lidar fournit des données altimétriques de la surface terrestre à une précision encore inégalée par les techniques stéréoscopiques classiques. Sous la forme d’un nuage de points tri-dimensionnel, nous présentons dans cet article une méthode pour générer un Modèle Numérique de Terrain à partir de ces données lidar conjointement avec des données image. Nous nous intéressons particulièrement au milieux ruraux peu urbanisés pour lesquels une cartographie grande échelle est un enjeu majeur. L’algorithme que nous proposons est basé sur un filtrage prédictif de Kalman pour lequel la composante temporelle est remplacée par une indexation spatiale. Appliqué au calcul de la pente locale et de l’altitude du terrain, il s’agit de combiner linéairement une «mesure» basée sur l’analyse du nuage de points dans un environnement local cylindrique et une « prédiction » basée sur les calculs déjà effectués. Le facteur de combinaison linéaire est calculé en fonction des incertitudes respectives sur la «mesure» et sur la « prédiction » des états du système. Nous soulignons également l’importance de la prise en compte de la pente locale pour la détermination de la hauteur du sol.

Si les données lidar fournissent parfois des informations altimétriques sur le terrain en présence de végétation, la densité de points au sol en présence de végétation dense devient très faible. Nous introduisons alors une adaptation du système de voisinage local basé sur l’intégration de données image et d’intensité lidar au sein d’un prédicteur de zones de végétation haute. Celui-ci s’accroît lorsque la densité de points au sol diminue, augmentant ainsi la probabilité de trouver des points sol. Nous présentons pour finir des résultats prometteurs pour la poursuite de ce travail sur le Golfe du Morbihan.

Keywords: 

Airborne Lidar, Digital Terrain Models, Vegetation Indice, Classification, Kalman filter.

Mots clés

Lidar aéroporté, Modèles Numériques de Terrain, Indice de végétation, Classification, Filtre de Kalman.

1. Introduction
2. Problématique
3. Génération de MNT par Filtrage Prédictif
4. Fusion Lidar/Image
5. Les Données : le Golfe du Morbihan
6. Résultats et Discussions
7. Conclusion
  References

[1] E. AHOKAS, S. KAASALAINEN, J. HYYPPA and J. SUOMALAINEN, Calibration of the Optech ALTM 3100 laser scanner intensity data using brightness targets, In Proc. of the ISPRS Commission I Symposium, IAPRS, Marne-la-Vallee, France, jul 2006.

[2] P. AXELSSON, Dem generation from laser scanner data using adaptative tin models, In International Archives of Photogrammetry and Remote Sensing, volume XXXIII part B4/1, pages 110-117, 2000.

[3] C. BRIESE, N. PFEIFER and P. DORNIGER, Applications of the robust interpolation for a dtm determination. In R.Kalliany and F. Leberl, editors, Proc. of the ISPRS Commission III Symposium on Photogrammetric and Computer Vision, volume XXXIV of The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, pages 55-61, Graz, Austria, Sep 2002. Institute for Computer Graphics and Vision, Graz University of Technology.

[4] A. CHAUVE, C. MALLET, F. BRETAR, S. DURRIEU, M. PIERROT-DESEILLIGNY and W. PUECH, Processing full-waveform lidar data : modeling raw signals, In Proc. of the ISPRS Workshop III/3 'LaserScanning 2007', Espoo, Finland, sep 2007. ISPRS.

[5] F. COREN and P. STERZAI, Radiometric correction in laser scanning, International Journal of Remote Sensing, 27(15) :3097-3104, August 2006.

[6] F. COREN, D. VISINTINI, PREARO G. and P. STERZAI, Integrating lidar intensity measures and hyperspectral data for extracting of cultural heritage, In Proc. of Workshop Italy-Canada for 3D Digital Imaging and Modeling : applications of heritage, industry, medicine and land., 2005.

[7] N. DAVID, C. MALLET and F. BRETAR, Library concept and design for lidar data processing, In Geographic Object Based Image Analysis, 2008.

[8] W. ECKSTEIN and O. MUNKELT, Extracting objects from digital terrain models, In Proc. Int. Society for Optical Engineering : Remote Sensing and Reconstruction for Three-Dimensional Objects and Scenes, volume 2572, pages 43-51, 1995.

[9] G. EVENSEN, Data assimilation, The ensemble Kalman Filter. Springer, oct 2006.

[10] B. HOE and N. PFEIFER, Correction of laser scanning intensity data : Data and model-driven approaches, ISPRS Journal of Photogrammetry and Remote Sensing, (in press), 2007.

[11] M. KASSER and Y. EGELS, Digital Photogrammetry, Hermes - Lavoisier, 2002.

[12] J. KILIAN, N. HAALA and M. ENGLICH, Capture and evaluation of airborne laser scanner data, In International Archives of Photogrammetry and Remote Sensing, volume XXXI, pages 383-388, 1996.

[13] K. KRAUS and N. PFEIFER, Determination of terrain models in wooded areas with airborne laser scanner data, ISPRS Journal of Photogrammetry and Remote Sensing, 53 :193-203, 1998.

[14] H.S. LEE and N.H. YOUNAN, Dtm extraction of lidar returns via adapta-tive processing, IEEE Transactions on Geoscience and Remote Sensing, 41(9) :2063-2069, sep 2003.

[15] T. LILLESAND and R. KIEFER, Remote Sensing and Image interpretation, JohnWiley & Sons, 1994.

[16] P. LOHMAN, A. KOCH and M. SCHAEFFER, Approaches to the filtering of laser scanner data. In T. Schenk and G. Vosselman, editors, Proc. of the XIXth ISPRS Congress, volume XXXIII of The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, pages 540-547, Amsterdam, The Netherlands, Jul 2000. ISPRS.

[17] B. PETZOLD, P. REISS and W. STOSSEL, Laser scanning-surveying and mapping agencies are using a new technique for the derivation of digital terrain models, ISPRS Journal of Photogrammetry and Remote Sensing, 54 :95-104, 1999.

[18] M. PIERROT-DESEILLIGNY and N. PAPARODITIS, A multiresolution and optimization-based image matching approach : An application to surface reconstruction from spot5-hrs stereo imagery. In Proc. of the ISPRS Conference Topographic Mapping From Space (With Special Emphasis on Small Satellites), Ankara, Turkey, feb 2006. ISPRS.

[19] G. SITHOLE and G. VOSSELMAN, Experimental comparison of filter algorithms for bare-earth extraction from airborne laser scanning point clouds, ISPRS Journal of Photogrammetry and Remote Sensing, 59(1-2) :85-101, 2004.

[20] W. WAGNER, A. ULLRICH, T. MELZER, C. BRIESE and K. KRAUS, From singlepulse to full-waveform airborne laser scanners : potential and practical challenges, In The International Archives of the Photogrammetry and Remote Sensing, volume Vol. XXXV, part B3, pages 201-206, Istanbul, Turkey, Jul 2004. ISPRS.

[21] G. XU and Z. ZHANG, Epipolar Geometry in stereo, motion and object recognition, Kluwer Academic Publishers, 1996.

[22] K. ZHANG, S-C. CHEN, D. WHITMAN, M. SHYU, J. YAN and C. ZHANG. A progressive morphological filter for removing nonground measurements from airborne lidar data, IEEE Transactions on Geoscience and Remote Sensing, 41(4) :872-882, apr 2003.