3D reconstruction of hollow parts analyzing images acquired by a fiberscope

被引:6
作者
Icasio-Hernandez, Octavio [1 ,2 ]
Gonzalez-Barbosa, Jose-Joel [1 ]
Hurtado-Ramos, Juan B. [1 ]
Viliesid-Alonso, Miguel [2 ]
机构
[1] Inst Politecn Nacl, CICATA Unidad Queretaro, Queretaro 76090, Mexico
[2] Ctr Nacl Metrol CENAM, Queretaro 76246, Mexico
关键词
fiberscope; honeycomb structure; ring gages; CMM; CAMERA CALIBRATION; ENHANCEMENT;
D O I
10.1088/0957-0233/25/7/075402
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
A modified fiberscope used to reconstruct difficult-to-reach inner structures is presented. By substituting the fiberscope's original illumination system, we can project a profile-revealing light line inside the object of study. The light line is obtained using a sandwiched power light-emitting diode (LED) attached to an extension arm on the tip of the fiberscope. Profile images from the interior of the object are then captured by a camera attached to the fiberscope's eyepiece. Using a series of those images at different positions, the system is capable of generating a 3D reconstruction of the object with submillimeter accuracy. Also proposed is the use of a combination of known filters to remove the honeycomb structures produced by the fiberscope and the use of ring gages to obtain the extrinsic parameters of the camera attached to the fiberscope and the metrological traceability of the system. Several standard ring diameter measurements were compared against their certified values to improve the accuracy of the system. To exemplify an application, a 3D reconstruction of the interior of a refrigerator duct was conducted. This reconstruction includes accuracy assessment by comparing the measurements of the system to a coordinate measuring machine. The system, as described, is capable of 3D reconstruction of the interior of objects with uniform and non-uniform profiles from 10 to 60 mm in transversal dimensions and a depth of 1000 mm if the material of the walls of the object is translucent and allows the detection of the power LED light from the exterior through the wall. If this is not possible, we propose the use of a magnetic scale which reduces the working depth to 170 mm. The assessed accuracy is around +/- 0.15 mm in 2D cross-section reconstructions and +/- 1.3 mm in 1D position using a magnetic scale, and +/- 0.5 mm using a CCD camera.
引用
收藏
页数:12
相关论文
共 40 条
[1]  
Albertazzi Jr G, 2008, P SOC PHOTO-OPT INS, V7155
[2]  
[Anonymous], 2002, B89162002 ASME
[3]  
Antunes M.A.A., 2010, THESIS U COIMBRA POR
[4]  
Arai K, 2012, INT J ADV RES ARTIF, V1
[5]  
Barrado JMD, 2009, BIBLIOTEC HIST, P1
[6]  
Bernhardt Sylvain, 2013, Medical Computer Vision. Recognition Techniques and Applications in Medical Imaging. Second International MICCAI Workshop, MCV 2012. Revised Selected Papers, P254, DOI 10.1007/978-3-642-36620-8_25
[7]  
Biegelbauer G., 2006, IEEE International Conference on Computer Vision Systems, 2006 ICVS '06, P22, DOI DOI 10.1109/ICVS.2006.1
[8]  
*BIPM, 2008, EV MEAS DATAGUIDE EX
[9]  
Bouguet J., 2010, CAMERA CALIBRATION T
[10]   Size measurement in endoscopic images by low coherence interferometry [J].
Criante, L. ;
Lucesoli, A. ;
Farabollini, B. ;
Bonifazi, F. ;
Rozzi, T. ;
Simoni, F. .
JOURNAL OF OPTICS A-PURE AND APPLIED OPTICS, 2009, 11 (03)