iccv iccv2013 iccv2013-17 iccv2013-17-reference knowledge-graph by maker-knowledge-mining

17 iccv-2013-A Global Linear Method for Camera Pose Registration


Source: pdf

Author: Nianjuan Jiang, Zhaopeng Cui, Ping Tan

Abstract: We present a linear method for global camera pose registration from pairwise relative poses encoded in essential matrices. Our method minimizes an approximate geometric error to enforce the triangular relationship in camera triplets. This formulation does not suffer from the typical ‘unbalanced scale ’ problem in linear methods relying on pairwise translation direction constraints, i.e. an algebraic error; nor the system degeneracy from collinear motion. In the case of three cameras, our method provides a good linear approximation of the trifocal tensor. It can be directly scaled up to register multiple cameras. The results obtained are accurate for point triangulation and can serve as a good initialization for final bundle adjustment. We evaluate the algorithm performance with different types of data and demonstrate its effectiveness. Our system produces good accuracy, robustness, and outperforms some well-known systems on efficiency.


reference text

[1] S. Agarwal, N. Snavely, and S. Seitz. Fast algorithms for l∞ problems in multiview geometry. In Proc. CVPR, pages 1–8, 2008. 2

[2] S. Agarwal, N. Snavely, I. Simon, S. Seitz, and R. Szeliski. Building rome in a day. In Proc. ICCV, 2009. 1, 2

[3] M. Arie-Nachimson, S. Z. Kovalsky, I. KemelmacherShlizerman, A. Singer, and R. Basri. Global motion estimation from point matches. In Proc. 3DPVT, 2012. 1, 2, 5, 6

[4] M. Brand, M. Antone, and S. Teller. Spectral solution of large-scale extrinsic camera calibration as a graph embedding problem. In Proc. ECCV, 2004. 2

[5] A. Buchanan and A. Fitzgibbon. Damped newton algorithms for matrix factorization with missing data. In Proc. CVPR, pages 316–322, 2005. 2

[6] P. Chen and D. Suter. Recovering the missing components in a large noisy low-rank matrix: application to sfm. IEEE Trans. PAMI, 26(8): 1051–1063, 2004. 2

[7] J. Courchay, A. Dalalyan, R. Keriven, and P. Sturm. Exploiting loops in the graph of trifocal tensors for calibrating a network of cameras. In Proc. ECCV, pages 85–99, 2010. 2

[8] D. Crandall, A. Owens, N. Snavely, and D. Huttenlocher. Discrete-continuous optimization for large-scale structure from motion. In Proc. CVPR, pages 3001–3008, 2011. 2

[9] A. Dalalyan and R. Keriven. L1-penalized robust estimation for a class ofinverse problems arising in multiview geometry. In NIPS, 2009. 2

[10] M. Fischler and R. Bolles. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Communications oftheACM, 24(6):381–395, 1981. 4

[11] A. Fitzgibbon and A. Zisserman. Automatic camera recovery for closed or open image sequences. Proc. ECCV, pages 311–326, 1998. 1, 2 487

[12] Y. Furukawa, B. Curless, S. M. Seitz, and R. Szeliski. Towards internet-scale multi-view stereo. In Proc. CVPR, 2010. 6

[13] V. M. Govindu. Combining two-view constraints for motion estimation. In Proc. CVPR, pages 218–225, 2001 . 1, 2, 3

[14] V. M. Govindu. Lie-algebraic averaging for globally consistent motion estimation. In Proc. CVPR, 2004. 2

[15] R. Hartley, J. Trumpf, Y. Dai, and H. Li. Rotation averaging. IJCV, pages 1–39, 2013. 2, 3, 4

[16] R. Hartley and A. Zisserman. Multiple View Geometry in Computer Vision. Cambridge University Press, 2003. 3

[17] M. Havlena, A. Torii, J. Knopp, and T. Pajdla. Randomized structure from motion based on atomic 3d models from camera triplets. In Proc. CVPR, pages 2874–2881, 2009. 1, 2

[18] N. Jiang, P. Tan, and L. Cheong. Seeing double without confusion: Structure-from-motion in highly ambiguous scenes.

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30] In Proc. CVPR, pages 1458–1465, 2012. 2 F. Kahl. Multiple view geometry and the l∞ norm. In Proc. ICCV, 2005. 1, 2 Q. Ke and T. Kanade. Robust l1 norm factorization in the presence of outliers and missing data by alternative convex programming. In Proc. CVPR - Volume 1, pages 739–746, 2005. 2 R. Lehoucq and J. Scott. An evaluation of software for computing eigenvalues of sparse nonsymmetric matrices. Preprint MCS-P547, 1195, 1996. 5 M. Lhuillier and L. Quan. A quasi-dense approach to surface reconstruction from uncalibrated images. IEEE Trans. PAMI, 27(3):418–433, 2005. 1, 2 H. Li and R. Hartley. Five-point motion estimation made easy. In Proc. ICPR, pages 630–633, 2006. 1 D. Martinec and T. Pajdla. Robust rotation and translation estimation in multiview reconstruction. In Proc. CVPR, pages 1–8, 2007. 1, 2, 3 D. Nist e´r. An efficient solution to the five-point relative pose problem. IEEE Trans. PAMI, 26:756–777, 2004. 1, 2, 4, 5 D. Nist e´r and F. Schaffalitzky. Four points in two or three calibrated views: Theory and practice. IJCV, 67(2):21 1–23 1, 2006. 1, 2, 5, 6 D. Nister and H. Stewenius. Scalable recognition with a vocabulary tree. In Proc. CVPR, 2006. 4 C. Olsson, A. Eriksson, and R. Hartley. Outlier removal using duality. In Proc. CVPR, pages 1450–1457, 2010. 2 C. Olsson, A. Eriksson, and F. Kahl. Efficient optimization for l∞ problems using pseudoconvexity. In Proc. ICCV, 2007. 2 M. Pollefeys, R. Koch, and L. Gool. Self-calibration and

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39] metric reconstruction inspite of varying and unknown intrinsic camera parameters. IJCV, 32(1):7–25, 1999. 1 M. Pollefeys, L. Van Gool, M. Vergauwen, F. Verbiest, K. Cornelis, J. Tops, and R. Koch. Visual modeling with a hand-held camera. IJCV, 59:207–232, 2004. 2 L. Quan. Invariants of six points and projective reconstruction from three uncalibrated images. IEEE Trans. PAMI, 17(1):34–46, 1995. 1, 2 R. Roberts, S. Sinha, R. Szeliski, and D. Steedly. Structure from motion for scenes with large duplicate structures. In Proc. CVPR, 2011. 2 P.-Y. Shen, W. Wang, C. Wu, L. Quan, and R. Mohr. From fundamental matrix to trifocal tensor. In Proc. SPIE, volume 3454, pages 340–347, 1998. 2 S. Sinha, D. Steedly, and R. Szeliski. A multi-stage linear approach to structure from motion. In ECCV Workshop on Reconstruction and Modeling of Large-Scale 3D Virtual Environments, 2010. 1, 2, 6 N. Snavely, S. Seitz, and R. Szeliski. Photo tourism: exploring photo collections in 3d. ACM Trans. on Graph., 25:835– 846, 2006. 1, 2, 6 C. Strecha, W. von Hansen, L. Van Gool, P. Fua, and U. Thoennessen. On benchmarking camera calibration and multi-view stereo for high resolution imagery. In Proc. CVPR, 2008. 6 P. F. Sturm and B. Triggs. A factorization based algorithm for multi-image projective structure and motion. In Proc. ECCV (2), pages 709–720, 1996. 2 C. Tomasi and T. Kanade. Shape and motion from image streams under orthography: A factorization method. IJCV, 9: 137–154, 1992. 2

[40] P. Torr and A. Zisserman. Robust parameterization and computation of the trifocal tensor. Image and Vision Computing, 15:591–605, 1997. 1, 2

[41] B. Triggs, P. Mclauchlan, R. Hartley, and A. Fitzgibbon. Bundle adjustment - a modern synthesis. Lecture Notes in Computer Science, pages 298–375, 2000. 1

[42] C. Wu. Visualsfm: A visual structure from motion system. 2011. 6, 7

[43] C. Wu, S. Agarwal, B. Curless, and S. Seitz. Multicore bundle adjustment. In Proc. CVPR, pages 3057–3064, 2011. 5

[44] C. Zach, A. Irschara, and H. Bischof. What can missing correspondences tell us about 3d structure and motion? In Proc. CVPR, 2008. 2

[45] C. Zach, M. Klopschitz, and M. Pollefeys. Disambiguating visual relations using loop constraints. In Proc. CVPR, 2010. 2, 4, 6

[46] C. Zach and M. Pollefeys. Practical methods for convex multi-view reconstruction. In Proc. ECCV: Part IV, pages 354–367, 2010. 1, 2 Appendix A. Derivation of Equation (3) We first show that the length of the line segments ciA, cjB are approx- siikj sjikj imately | |ci − cj | | and | |ci − cj | | respectively. The three vector|s| cij , cik a|n adn cjk sh|o|culd− b ce |c|lo rsees ptoe coplanar, so the angle ∠Acick is close to zero, and the length of ciA is close to that of cick. We can calculate the length of cick as: ssiinn((θθjk))||ci− cj|| ≈ssiinn((θθk?j?))||ci− cj|| = siijk||ci− cj||. Note that θj? ≈ θj , θk? ≈ θk because the three vectors cij , cik and cjk are ≈clo θse to coplanar. The 3D coordinate of A is then approximated by ci + siikj | |ci − cj | |cik. Similarly, we sjikj can obtain the coordinate of B as cj + | |ci −cj | |cjk. As a result, the coordinate of ck, which is the midpoint ocf AB, can be computed by Equation (3). 488