Trois Approaches de Planification de Vue Automatiques et Intelligentes pour la Numérisation 3D d’Objetsinconnus

Trois Approaches de Planification de Vue Automatiques et Intelligentes pour la Numérisation 3D d’Objetsinconnus

Souhaiel Khalfaoui Ralph Seulin  Yohan Fougerolle  David Fofi 

Vecteo SAS IUT, 12 rue de la fonderie, 71200 Le Creusot, France

Laboratoire d’Electronique, Informatique et Image, IUT, 12 rue de la fonderie, 71200 Le Creusot, France

Page: 
245-269
|
DOI: 
https://doi.org/10.3166/TS.31.245-269
Received: 
4 October 2013
| |
Accepted: 
12 May 2014
| | Citation

OPEN ACCESS

Abstract: 

This paper presents three methods for the digitization of 3D objects without prior knowledge on their shape. The first method is simple and naïve and is based on the generation of view points by sampling the bounding box of the acquired data at each step of the acquisition process. The second method is an analysis of the orientation of the scanned parts. The third method explores the barely visible surfaces and is a combination of the angular visibility and the real one by ray tracing. Tests with objects of different complexity classes were performed. The results of digitization are provided and prove the efficiency and the robustness of our approaches. 

Extended Abstract

Context

The 3D models of objects are widely used for an increasing number of applications such as industrial applications, entertainment, preservation of important cultural heritage artefacts, and architectural applications. In industrial applications, objectsare digitized for inspection tasks, reverse engineering, and replication. Such applications demand high quality and accurate 3D models. The manual 3D digitization process is expensivesince itrequires ahighly trained technician who decides aboutthe different views needed to acquire the object model. The quality of the final result strongly depends, in addition to the complexity of the object shape, on the selected viewpoints and thus on the human expertise. Thus, this technique does not fulfill the high level requirement of industrial applications which require reliable, repeatable, and fast programming routines. Therefore, it is necessary to develop an efficient automatic digitization strategy while minimizing the impact of the human factor.

The automatic scanning deploys the most beneficial views for the reconstruction process to achieve the highest possible accuracy and coverage rate with the smallest number of views. The automation problem becomes a View Planning problem in which one seeks for the Next Best Views (NBVs) to improve an existing model reconstruction and to cover all the object surface. The non-model based NBV planning can be seen as an incremental approach to building object or scene models using all the previously acquired 3D data.

Methods

This paper presents three view planning approaches for the digitization of 3D objects without prior knowledge on their shape. The first method is simple and naïve and is based on the generation of view points by sampling the bounding box of the acquired data at each step of the acquisition process.

The second method, called Orientation Clustering (OC), is an analysis of the orientation of the scanned parts and uses the concept of Mass Vector Chains (Yuan, 1995) to define the global orientation of the scanned part. All of the viewpoints satisfying an orientation constraint are clustered using the Mean Shift technique (Comaniciu, Meer, 2002) to construct a first set of candidates for the NBV. Then, a weight is assigned to each mode according to the elementary orientations of its different descriptors. The NBV is chosen among the modes with the highest weights and which comply with the robotics constraints.

The third method is called Barely Visible Surfaces Clustering (BVSC) and is mainly composed of two stages: the target points identification and the viewpointsbility definition and the ray tracing test (Whitted, 1980) to explore the barely visible surfaces. The viewpoints are then selected according to a clustering step with Mean Shift technique to determine a set of NBVs.

RÉSUMÉ

Ce papier propose trois approches itératives et intelligentes de planification de vue pour la numérisation 3D d’objets sans connaissance a priori de leurs formes. La première méthode est une approche simple et naïve basée sur la génération d’un ensemble de points de vue par échantillonnage régulier de l’enveloppe englobante des données acquises. La deuxième méthode est basée sur une analyse de l’orientation des différentes parties acquises. La troisième méthode vise à explorer les parties de l’objet qui figurent dans la limite du champ de visibilité et est basée sur un couplage de la visibilité angulaire avec la visibilité réelle par lancer de rayons. Les résultats de numérisation d’objets de différentes classes de complexité sont présentés et prouvent l’efficacité et la robustesse de nos approches. 

Keywords: 

view planning, 3D digitization, automation, non model-based method.

MOTS-CLÉS

planificationdeprisedevues,numérisation3D,automatisation,méthodenon-basée sur un modèle.

1. Introduction
2. Travaux Antérieurs
3. Méthodes Proposées
4. Simulations et Discussion
5. Expérimentation et Résultats
6. Conclusions
  References

Abidi B. (1995). Automatic sensor placement. In SPIE Conference on Intelligent Robots and Computer Vision XIV: Algorithms, Techniques, Active Vision, and Materials Handling, vol. 2588, p. 387-398. Philadelphia, PA. 

Chen S., Li Y. F., Zhang J., Wang W. (2008). Active sensor planning for multiview vision tasks (1st éd.). Springer Publishing Company. 

Comaniciu D., Meer P. (2002). Mean shift: A robust approach toward feature space analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, p. 603-619. 

ConnollyC.I. (1985). Thedeterminationofnextbestviews. InProceedingsoftheInternational Conference on Robotics and Automation, p. 432-435. St. Louis, Mo., USA. 

Dunn E., Frahm J.-M. (2009). Next best view planning for active model improvement. In Proceedings of the British Machine Vision Conference, p. 53.1-53.11. 

FoissotteT.,StasseO.,EscandeA.,KheddarA. (2008). Anext-best-viewalgorithmforautonomous 3D object modeling by a humanoid robot. In International Conference on Humanoid Robots, p. 333–338.

Imaging, Robotics, and Intelligent Systems Laboratory. (2003). The 3D Database, http://imaging.utk.edu. 

Khalfaoui S., Aigueperse A., Seulin R., Fougerolle Y., Fofi D. (2012). Fully automatic 3D digitization of unknown objects using progressive data bounding box. In Proceedings of SPIE Electronic Imaging, p. 829011-829011-8. San Francisco. 

Khalfaoui S., Seulin R., Fougerolle Y., Fofi D. (2012). View planning approach for automatic 3D digitization of unknown objects. In Proceedings of the European Conference on Computer Vision, p. 496-505. 

Kriegel K., Bodenmüller T., Suppa M., Hirzinger G. (2011). A surface-based next-best-view approach for automated 3D model completion of unknown objects. In Proceedings of the International Conference on Robotics and Automation (ICRA), p. 4869-4874.

Loriot B., Seulin R., Gorria P. (2008). Non-model based method for an automation of 3D acquisitionandpost-processing. Electronics Letters on Computer Vision and Analysis,vol.7, p. 67-82. 

Low K.-L., Lastra A. (2006). Efficient constraint evaluation algorithms for hierarchical nextbest-viewplanning. InInternationalSymposiumon3DDataProcessing,Visualization,and Transmission, p. 830–837. Washington, DC, USA. 

MassiosN.A.,FisherR.B. (1998). Abestnextviewselectionalgorithmincorporatingaquality criterion. In British Machine Vision Conference (BMVC), p. 780-789. Southampton, UK. 

Maver J., Bajcsy R. (1993). Occlusions as a guide for planning the next view. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 15, no 5, p. 417–433. 

Munkelt C., Denzler J., Kuhmstedt P. (2006). Incorporation of a-priori information in planning the next-best-view. In International Workshop on Vision, Modeling, and Visualization (VMV), p. 261-268. Aachen. 

Olague G., Mohr R. (2002, April). Optimal camera placement for accurate reconstruction. Pattern Recognition, vol. 35, p. 927-944. 

Pito R. (1999, octobre). A solution to the next best view problem for automated surface acquisition. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 21, no 10, p. 1016–1030. 

Scott W. R., Roth G., Rivest J.-F. (2003). View planning for automated three-dimensional object reconstruction and inspection. ACM Computing Surveys, vol. 35, p. 64-96. 

Sukumar S. R., Koschan A. F., Abidi M. A. (2008, June). Towards understanding what makes 3Dobjectsappearsimpleorcomplex. InIEEEConferenceonComputerVisionandPattern Recognition, p. 1-8. 

Whitted T. (1980). An improved illumination model for shaded display. Communications of the ACM, vol. 23, p. 343-349. 

YuanX. (1995). Amechanismofautomatic3Dobjectmodeling. IEEETransactionsonPattern Analysis and Machine Intelligence, vol. 17, no 3, p. 307–311.