G. Sansoni, S. Carmignato, E. Savio: Validation of the measurement performance of a three dimensional vision sensor by means of a Coordinate Measurement Machine
IMTC 2004
Validation of the measurement performance of a three dimensional vision sensor by means of a Coordinate Measurement Machine Giovanna Sansoni 1, Simone Carmignato 2, Enrico Savio2 1 Laboratory of Optoelectronics, University of Brescia Via Branze, 38, 25123 Brescia BS, Italy 2 Laboratory of Industrial and Geometrical Metrology, University of Padova Via Venezia, 1, 35131 Padova PD, Italy Correspondence to Giovanna Sansoni,
[email protected]
Keywords Optical digitisation, metrological validation, CMM.
Introduction This paper presents the results of the activity carried out to perform the metrological validation of a three-dimensional (3-D) non-contact digitiser based on optical active triangulation. The system, named OPL-3D, is one of the deliverables of the project “Development of a novel methodology for the reverse engineering of complex, free-form surfaces, combining three-dimensional vision systems and Coordinate Measuring Machines”, funded in 2000 by the Italian Ministry of Research. The research activity was focused at the development of procedures that combine the measurement information from a 3-D vision sensor and a touch probe Coordinate Measuring Machine (CMM) for the Reverse Engineering (RE) of free-form surfaces. The objective was to reconstruct the CAD model of complex shapes with high accuracy and at the same time rapidly, and minimising the operator time [1]. This aspect is of great importance in the manufacturing industry, where the optimisation of the RE process in terms of both the accuracy and the efficiency is mandatory [2,3]. The optical system developed for the noncontact acquisition of the shapes is shown in Figure 1. The optical head is composed of a projector and of a video camera, mounted on a tripod. They are oriented following a typical triangulation approach [4]. The projector projects a sequence of eleven fringe patterns of incoherent light onto the object, following the well-known Gray Code-Phase Shift method. The video camera captures the patterns. These ones appear deformed by the object shape, Figure 1: OPL-3D due to the triangulation geometry of the system. Suitable developed procedures extract the shape of the object
1
G. Sansoni, S. Carmignato, E. Savio: Validation of the measurement performance of a three dimensional vision sensor by means of a Coordinate Measurement Machine
from the pattern deformation in the form of dense point clouds, retaining either the (x,y,z) point coordinates and the intensity (Gray scale coded) or the colour (in the RGB standard). OPL-3D is easy movable in space: the measurement of complex surfaces is performed by moving the system around the object, and by acquiring a number of partial views sufficient to gauge the whole object. Specifically designed software tools estimate the rototranslation matrix between the point clouds and align them in a common reference system. Specialised software, largely used in Reverse Engineering applications, transforms the point cloud into suitable 3D models (both triangle meshes and CAD surfaces) that feed the CMM, and significantly simplify the contact digitisation task. As part of the work carried out, the CMM was also used to perform the metrological validation of OPL-3D and of the view registration software. In fact, we thought very important to verify the quality of the point clouds by using an instrument with lower measuring uncertainty with respect to that one provided by the optical sensor, also in view of using it as the only source of measurement in those applications where the fast execution of the RE process is more important than its accuracy and whole-optical RE processes can be fruitfully exploited [5,6].
Proposed approach Target of the measurement The proposed approach for the performance verification of the 3D vision sensor is a task specific approach [7], which can be used to evaluate the accuracy of measurements for virtually any measurement task. For the procedure validation, experimental tests were performed on the turbine blade of Figure 2 (the deformation induced by the object shape on one of the fringes patterns is shown).
Figure 2: The turbine blade used as test object.
Figure 3: The point clouds acquired by OPL-3D. 3.a: single view; 3.b: multi-view acquisition.
This object (180 mm x 90 mm x 20 mm) was measured in two different set-ups: the former allowed us to capture the whole object in a single view; the latter resulted into the acquisition of three different views, partially overlapped. Figure 3 visualises the point clouds corresponding to the single view acquisition (left image) and to the multi-view acquisition
2
G. Sansoni, S. Carmignato, E. Savio: Validation of the measurement performance of a three dimensional vision sensor by means of a Coordinate Measurement Machine
(right image). The black holes represent non-measured portions of the surface, where proper marker had been placed to optimise the performance of the alignment [8,9]. Metrological validation The validation of the point clouds accuracy was performed by the procedure schematically shown in Figure 4. The CMM is used as the link to the traceability chain, to obtain an accurate CAD model of the measured physical object. Physical object CMM digitization
Optical digitization
Point cloud
Accurate CAD surfaces
CMM uncertainty
CAD – Cloud deviation
Accuracy evaluation of the optical system
Figure 4: Measurement of the turbine blade on CMM.
To measure the turbine blade, a CMM with Maximum Permissible Error (MPE) [10] of 2.2 + L/300 µm (L in mm) was used. The CMM was equipped with dedicated software to measure free-form surfaces. The probing configuration – which is shown in Figure 5 – was used in scanning contact mode.
Figure 5: Measurement of the turbine blade on CMM.
The procedure was tested and validated with reference to the turbine blade case study. Experimental results showed an accuracy of 0.1 mm
3
G. Sansoni, S. Carmignato, E. Savio: Validation of the measurement performance of a three dimensional vision sensor by means of a Coordinate Measurement Machine
(95% of points) when single view acquisition is considered. This value increases to 0.3 mm (95% of points), for the multiple-view point cloud. In the paper, the validation procedure will be detailed. Moreover, it will be shown how both the orientation of the fringes with respect to the object and the alignment software influence the measurement performance. Further experimental cases applied to the RE of free-form shapes will be presented: they will highlight that OPL-3D can be successfully used for reproduction of replicas as well as for the fast generation of CAD, NURBSbased models.
References [1] V. Carbone, M. Carocci, E. Savio, G. Sansoni, L. De Chiffre, "Combination of a vision system and a coordinate measuring machine for the reverse engineering of freeform surfaces", Advanced Manufacturing Technology, vol. 17, pp. 263-271, 2001.[2] [2] L. C. Chen, G. C. I. Lin, “An integrated reverse engineering approach to reconstructing free-form surfaces,” Comp. Integr. Manuf. Syst., Vol. 10, No. 1, pp. 4960, 1997. [3] L. C. Chen, G. C. I. Lin., “A vision-aided reverse engineering approach to reconstructing free-form surfaces,” Robotic & Computer-Integrated Manufacturing, Vol. 13, No. 4, pp. 323-336, 1997. [4] G. Sansoni, A. Patrioli and F. Docchio, "OPL-3D: a novel, portable optical digitiser for fast acquisition of free-form surfaces," Rev. Scient. Instr.,Vol. 74, N. 4, pp. 25932603, 2003. [5] M. Soucy, G. Godin, R. Baribeau, F. Blais and M. Rioux, "Sensors and algorithms for the construction od digital 3-D colour model of real object," Proc. of Int. Conf. On Image Proc., Vol. 2, pp. 409-412, 1996. [6] W. Niem and J. Wingbermühle, "Automatic reconstruction of 3D objects using a mobile manoscopic camera," Proc. of Intern. Conf. on Recent Advances in 3-D Digital and Modeling, pp. 173-180, 1997. [7] Wilhelm, R.G., Hocken, R., Schwenke, H., 2001, Task Specific Uncertainty in Coordinate Measurement, Annals of the CIRP, 50/2: 553-563. [8] G. Sansoni, A. Patrioli. "Registration of multiple range views from a portable optical digitizer", Proc. of ODIMAPIII, Optoelectronic Distance Measurements and Applications, 405-410, Pavia, 2001. [9] Y. Chen, G. Medioni, “Object modelling by registration of multiple range images,” Image Vision and Computing, n.10, p. 145-155, 1992. [10] ISO 10360-2: 2001, Geometrical Product Specifications (GPS) – Acceptance test and reverification test for CMMs – Part 2: CMMs used for measuring size.
4
