Future Work

5.2.1 The Camera Tracker

The limitation and effect of the existing camera tracker is presented in Section 4.3.

Actually, the proposed algorithm can not fully work if the camera tracker limitation is not

45

essential problem for all camera trackers. The problem should be a defect of the existing camera tracking algorithms. Hence, modifying the existing algorithm or designing a new algorithm for the shooting condition mentioned in this thesis is required.

The typical structure of a camera tracker is shown in Figure 5.2.1. The problem mentioned is caused by the feature point tracking block in the camera tracker. To deal with the problem, the feature point tracking algorithm must be improved to be able to determine whether a tracked point is appearing or disappearing in some frame. With the improvement, a point is tracked only when it is visible to the camera, not through out the image sequence. To achieve the requirement, the feature point tracking algorithm must track feature points based on not only the low level information such as edge, color, and texture, but also the information of higher level like geometry relationship. Hence, algorithms of higher information processing like image interpretation and understanding might be integrated into the point tracking algorithm. Figure 5.2.1 System structure for depth-map recovery algorithm

Besides the limitation of camera shooting angle, another minor improvement for the feature point tracker is also required for better background removal performance. As mentioned in Section 4.2, some important features including edges and corners of the object are not tracked by the feature point tracker. The lack of certain points makes the following

process, the object mask generation in Section 3.3, becoming more complex and difficult to recover the shape of the object. In other words, more feature points provided, more accuracy the object mask becomes. Hence, the algorithm of feature point extraction could be improved to obtain more feature points on the object surface to support the background removal algorithm.

5.2.2 The Background Removal Algorithm

The background removal algorithm is composed of two steps. Currently, each step implements simple algorithm and works as a prototype. For the foreground/background separation step, the current algorithm applies a modified nearest neighbor algorithm on 3D point positions to separate points. The algorithm is fast and simple but not accurate enough.

The separation step should imports information, such as texture, edge, and shape segment, from the object image to assist the separation algorithm. These information could be useful to eliminate points which are near but do not belong to the object.

On the other hand, the object mask generation step must be improved to be capable of dealing with concave contour and holes. From Section 4.2, it is clear that the ability is crucial for the quality of reconstructed model. However, it is difficult to determine the concave contour and holes of the object even with sufficient 2D point information. To obtain the accuracy object mask, information including texture and edges from object images must be taken into consideration to the object mask generation, same as the previous step.

5.2.3 The Texture Mapping Block

Though the texture mapping is a well-developed algorithm, the algorithm is much complex than octree algorithm and takes much time to implement. Hence, the texture mapping block is not implemented in the current system. However, a 3D model reconstruction system is incomplete without the texture mapping process, as a 3D model is incomplete without texture. The texture mapping block must be implemented in the future.

47

Bibliography

[1] Anderson Technologies LLC, "SynthEyes 2007 Demo," in http://www.ssontech.com/, 2007.

[2] Autodesk, "Maya Personal Learning Edition 8.5," in

http://usa.autodesk.com/adsk/servlet/index?siteID=123112&id=7639525.

[3] P. A. Beardsley, P. H. S. Torr, and A. Zisserman, "3D Model Acquisition from Extended Image Sequences," in ECCV '96 : Proc. 4th European Conference on Computer Vision-Volume II, 1996, pp. 683-695.

[4] E. Boyer, "Object Models from Contour Sequences," in ECCV '96 : Proc. 4th European Conference on Computer Vision-Volume II, London, UK, 1996, pp.

109-118.

[5] M. S. Brown and W. B. Seales, "Beyond 2D images: effective 3D imaging for library materials," in DL '00: Proceedings of the fifth ACM conference on Digital libraries, New York, NY, USA, 2000, pp. 27-36.

[6] J. C. Carr, R. K. Beatson, J. B. Cherrie, T. J. Mitchell, W. R. Fright, B. C. McCallum, and T. R. Evans, "Reconstruction and representation of 3D objects with radial basis functions," in SIGGRAPH '01: Proceedings of the 28th annual conference on Computer graphics and interactive techniques, New York, NY, USA, 2001, pp. 67-76.

[7] CGAL Open Source Project, "Chapter 24 : 2D Alpha Shapes," 2006.

[8] C. M. Christoudias, B. Georgescu, and P. Meer, "Synergism in Low Level Vision,"

Porc. of 16th International Conference on Pattern Recognition, vol. 04, pp. 150-155, 2002.

[9] T. H. Cormen, C. E. Leiserson, R. L. Rivest, and C. Stein, Introduction to Algorithms, Second Edition: {The MIT Press}, 2001.

[10] B. Curless and M. Levoy, "A volumetric method for building complex models from range images," in SIGGRAPH '96: Proceedings of the 23rd annual conference on Computer graphics and interactive techniques, New York, NY, USA, 1996, pp.

303-312.

[11] P. Eisert, E. Steinbach, and B. Girod, "Automatic reconstruction of stationary 3-D objects from multiple uncalibrated camera views," Circuits and Systems for Video Technology, IEEE Transactions on, vol. 10, pp. 261-277, 2000.

[12] P. F. Felzenszwalb and D. P. Huttenlocher, "Efficient Graph-Based Image Segmentation," Int. J. Comput. Vision, vol. 59, pp. 167-181, September 2004.

[13] K. Fischer, "Introduction to alpha shapes," 2000.

[14] A. W. Fitzgibbon, G. Cross, and A. Zisserman, "Automatic 3D Model Construction for Turn-Table Sequences," in Proceedings of SMILE Workshop on Structure from Multiple Images in Large Scale Environments; Lecture Notes in Computer Science, 1998, pp. 154-170.

[15] S. F. Frisken and R. N. Perry, "Efficient estimation of 3D Euclidean distance fields from 2D range images," in VVS '02: Proceedings of the 2002 IEEE symposium on Volume visualization and graphics, Piscataway, NJ, USA, 2002, pp. 81-88.

[16] B. Georgescu and C. M. Christoudias, "Edge Detection and Image SegmentatiON (EDISON) System," in http://www.caip.rutgers.edu/riul/research/robust.html, 2003.

[17] S. Gibson, J. Cook, T. Howard, R. Hubbold, and D. Oram, "ICARUS v2.09 personal edition," 2003.

[18] K. Higuchi, M. Hebert, and K. Ikeuchi, "Building 3-D models from unregistered range images," Graph.Models Image Process., vol. 57, pp. 315-333, 1995.

[19] M. Kass, A. Witkin, and D. Terzopoulos, "Snakes: Active contour models," Int'l Conf.

on Computer Vision, pp. 321-331, 1987.

[20] R. Koch, M. Pollefeys, and L. J. V. Gool, "Multi Viewpoint Stereo from Uncalibrated Video Sequences," in ECCV '98 : Proc. 5th European Conference on Computer Vision-Volume I, London, UK, 1998, pp. 55-71.

[21] Laboratorium für Informationstechnologie University of Hannover, "voodoo camera tracker 0.9.1 beta," in http://www.digilab.uni-hannover.de/docs/manual.html, 2007.

[22] M. Levoy, K. Pulli, B. Curless, S. Rusinkiewicz, D. Koller, L. Pereira, M. Ginzton, S.

Anderson, J. Davis, J. Ginsberg, J. Shade, and D. Fulk, "The digital Michelangelo project: 3D scanning of large statues," in SIGGRAPH '00, New York, NY, USA, 2000, pp. 131-144.

[23] W. E. Lorensen and H. E. Cline, "Marching cubes: A high resolution 3D surface construction algorithm," in SIGGRAPH '87: Proceedings of the 14th annual conference on Computer graphics and interactive techniques, New York, NY, USA, 1987, pp. 163-169.

[24] C. Montani, R. Scateni, and R. Scopigno, "A modified look-up table for implicit disambiguation of Marching Cubes," The Visual Computer, vol. 10, pp. 353-355, December 1994.

[25] A. Moreira and M. Santos, "Concave Hull: a k-nearest neighbours approach for the computation of the region occupied by a set of points," 2007, pp. 8-11.

[26] C. Pheatt, J. Ballester, and D. Wilhelmi, "Low-cost three-dimensional scanning using range imaging," J.Comput.Small Coll., vol. 20, pp. 13-19, 2005.

[27] M. Pollefeys, R. Koch, M. Vergauwen, and L. V. Gool, "Flexible acquisition of 3D structure from motion," in Proc. 10th IMDSP Workshop, 1998, pp. 90-95.

[28] L. Qingquan, L. Bijun, L. Yuguang, and C. Jing, "3D Modeling and Visualization Based on Laserscanning," Geographic Information Sciences, vol. 6, pp. 159-164, 12 2000.

[29] J. Sethian, Level Set Methods and Fast Marching Methods: Evolving Interfaces in

49

[30] S. Sullivan and J. Ponce, "Automatic Model Construction, Pose Estimation, and Object Recognition from Photographs Using Triangular Splines," in ICCV '98:

Proceedings of the Sixth International Conference on Computer Vision, Washington, DC, USA, 1998, p. 510.

[31] B. Sumengen and B. S. Manjunath, "Graph Partitioning Active Contours (GPAC) for Image Segmentation," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, p. 509, 2006.

[32] B. Sumner, "Active Contour Models: Snakes."

[33] R. Szeliski, "Rapid octree construction from image sequences," CVGIP: Image Underst., vol. 58, pp. 23-32, 1993.

[34] The Pixel Farm, "PFTrack 4.0 evalution," 2007.

[35] N. Wolfgang and W. Jochen, "Automatic Reconstruction of 3D Objects using a Mobile Monoscopic Camera," in 3DIM '97 : Proc. Int. Conf. on Recent Advances in 3-D Digital Imaging and Modeling, Washington, DC, USA, 1997, p. 173.

[36] C. Xu and J. L. Prince, "Snakes, Shapes, and Gradient Vector Flow," IEEE Transactions on Image Processing, vol. 7, pp. 359-369, March 1998.

[37] Y. Yemez and F. Schmitt, "3D reconstruction of real objects with high resolution shape and texture," Image and Vision Computing, vol. 22, pp. 1137-1153, NOV 1 2004.

[38] C. T. Zahn, "Graph-theoretic methods for detecting and describing gestalt clusters,"

IEEE Transactions on Computing, vol. 20, pp. 68-86, 1971.