Abstract
Two-view relative pose estimation and structure reconstruction is a classical problem in computer vision. The typical methods usually employ the singular value decomposition of the essential matrix to get multiple solutions of the relative pose, from which the right solution is picked out by reconstructing the three-dimension (3D) feature points and imposing the constraint of positive depth. This paper revisits the two-view geometry problem and discovers that the two-view imaging geometry is equivalently governed by a Pair of new Pose-Only (PPO) constraints: the same-side constraint and the intersection constraint. From the perspective of solving equation, the complete pose solutions of the essential matrix are explicitly derived and we rigorously prove that the orientation part of the pose can still be recovered in the case of pure rotation. The PPO constraints are simplified and formulated in the form of inequalities to directly identify the right pose solution with no need of 3D reconstruction and the 3D reconstruction can be analytically achieved from the identified right pose. Furthermore, the intersection inequality also enables a robust criterion for pure rotation identification. Experiment results validate the correctness of analyses and the robustness of the derived pose solution/pure rotation identification and analytical 3D reconstruction.
Similar content being viewed by others
References
Bazin, J. C., Demonceaux, C., Vasseur, P., & Kweon, I. S. (2010). Motion estimation by decoupling rotation and translation in catadioptric vision. Computer Vision and Image Understanding, 114, 254–273.
Beardsley, P. A., & Zisserman, A. (1995). Affine calibration of mobile vehicles. In Europe-China Workshop on Geometrical Modelling and Invariants for Computer Vision, pp. 214–221.
Faugeras, O. D., & Maybank, S. (1990). Motion from point matches: Multiplicity of solutions. International Journal of Computer Vision, 4, 225–246.
Ferraz, L., Binefa, X., & Moreno-Noguer, F. (2014). Very fast solution to the PnP problem with algebraic outlier rejection. In Computer Vision and Pattern Recognition, pp. 501–508.
Garro, V., Crosilla, F., & Fusiello, A. (2012). Solving the PnP problem with anisotropic orthogonal procrustes analysis. In Second International Conference on 3d Imaging, Modeling, Processing, Visualization and Transmission, pp. 262–269.
Golub, G. H., & Van Loan, C. F. (1996) Matrix computations (3rd ed.). Johns Hopkins University Press.
Hartley, R. I. (1992). Estimation of relative camera positions for uncalibrated cameras. Presented at the ECCV ‘92 Proceedings of the Second European Conference on Computer Vision.
Hartley, R. I. (1995). An investigation of the essential matrix. Report Ge.
Hartley, R., Gupta, R., & Chang, T. (1992). Stereo from uncalibrated cameras. In Proceedings CVPR ‘92., 1992 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 761–764.
Hartley, R., & Zisserman, A. (2003). Multiple view geometry in computer vision. Cambridge: Cambridge University Press.
Horn, B. K. P. (1990). Relative orientation. International Journal of Computer Vision, 4, 59–78.
Huang, T. S., & Faugeras, O. D. (1989). Some properties of the E matrix in two-view motion estimation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 11, 1310–1312.
Huang, T. S. & Shim, Y. S. (1988). Linear algorithm for motion estimation: How to handle degenerate cases. In International Conference on Pattern Recognition, pp. 439–447.
Kneip, L., Siegwart, R., & Pollefeys, M. (2012). Finding the exact rotation between two images independently of the translation. In European Conference on Computer Vision, pp. 696–709.
Ling, L., Cheng, E., & Burnett, I. S. (2011). Eight solutions of the essential matrix for continuous camera motion tracking in video augmented reality. In Proceedings of the 2011 IEEE International Conference on Multimedia and Expo (ICME 2011), Barcelona, Spain, pp. 1–6.
Longuet-Higgins, H. C. (1981). A computer algorithm for reconstructing a scene from two projections. Nature, 293, 133–135.
Ma, Y., Soatto, S., Kosecka, J., & Sastry, S. S. (2004). An invitation to 3-D vision. Berlin: Springer.
Maybank, S. (1993). Theory of reconstruction from image motion (Vol. 28). Berlin: Springer.
Nister, D. (2004). An efficient solution to the five-point relative pose problem. IEEE Transactions on Pattern Analysis and Machine Intelligence, 26, 756–777.
Stewénius, H., Engels, C., & Nistér, D. (2006). Recent developments on direct relative orientation. ISPRS Journal of Photogrammetry & Remote Sensing, 60, 284–294.
Vieville, T., & Lingrand, D. (1996). Using singular displacements for uncalibrated monocular visual systems. In European Conference on Computer Vision, pp. 207–216.
Wang, W., & Tsui, H. T. (2000). An SVD decomposition of essential matrix with eight solutions for the relative positions of two perspective cameras. In International Conference on Pattern Recognition, Vol. 1, pp. 362–365.
Zhang, Z. (1998). Determining the epipolar geometry and its uncertainty: A review. International Journal of Computer Vision, 27, 161–195.
Acknowledgements
Thanks to anonymous reviewers for their constructive comments and Dr. Danping Zou for group talks. The work is funded by National Natural Science Foundation of China (61422311, 61673263, 61503403) and Hunan Provincial Natural Science Foundation of China (2015JJ1021).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by M. Hebert.
Appendix
Appendix
Cross Product
The cross product of two vectors a and b is defined only in three-dimensional space and is denoted by \( {\mathbf{a}} \times {\mathbf{b}} \). The vector triple product is the cross product of a vector with the result of another cross product, and is related to the dot product by the following formula
and
Kronecker Product
If A is an \( m \times n \) matrix and B is a \( p \times q \) matrix, then the Kronecker product \( A \otimes B \) is the \( mp \times nq \) block matrix:
Vectorization
the vectorization of an \( m \times n \) matrix A, denoted \( vec\left( A \right) \), is the \( mn \times 1 \) column vector obtained by stacking the columns of the matrix A on top of one another:
Here, \( a_{ij} \) represents the element of row-i and column-j. The vectorization is frequently used together with the Kronecker product to express matrix multiplication as a linear transformation on matrices. In particular,
for matrices A, B, and C of dimensions \( k \times l \), \( l \times m \), and \( m \times n \).
Rights and permissions
About this article
Cite this article
Cai, Q., Wu, Y., Zhang, L. et al. Equivalent Constraints for Two-View Geometry: Pose Solution/Pure Rotation Identification and 3D Reconstruction. Int J Comput Vis 127, 163–180 (2019). https://doi.org/10.1007/s11263-018-1136-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11263-018-1136-9