Optimization Criteria, Sensitivity and Robustness of Motion and Structure Estimation
The prevailing efforts to study the standard formulation of motion and structure recovery have been recently focused on issues of sensitivity and robustness of existing techniques. While many cogent observations have been made and verified experimentally, many statements do not hold in general settings and make a comparison of existing techniques difficult. With an ultimate goal of clarifying these issues we study the main aspects of the problem: the choice of objective functions, optimization techniques and the sensitivity and robustness issues in the presence of noise.
We clearly reveal the relationship among different objective functions, such as “(normalized) epipolar constraints”, “reprojection error” or “triangulation”, which can all be be unified in a new “ optimal triangulation” procedure formulated as a constrained optimization problem. Regardless of various choices of the objective function, the optimization problems all inherit the same unknown parameter space, the so called “essential manifold”, making the new optimization techniques on Riemanian manifolds directly applicable.
Using these analytical results we provide a clear account of sensitivity and robustness of the proposed linear and nonlinear optimization techniques and study the analytical and practical equivalence of different objective functions. The geometric characterization of critical points of a function defined on essential manifold and the simulation results clarify the difference between the effect of bas relief ambiguity and other types of local minima leading to a consistent interpretations of simulation results over large range of signal-to-noise ratio and variety of configurations.
KeywordsObjective Function Noise Level Motion Estimation High Noise Level Linear Algorithm
Unable to display preview. Download preview PDF.
- 1.K. Danilidis. Visual Navigation, chapter ”Understanding Noise Sensitivity in Structure from Motion”. Lawrence Erlbaum Associates, 1997.Google Scholar
- 3.A. Edelman, T. Arias, and S. T. Smith. The geometry of algorithms with orthogonality constraints. SIAM J. Matrix Analysis Applications, to appear.Google Scholar
- 6.A. D. Jepson and D. J. Heeger. Linear subspace methods for recovering translation direction. Spatial Vision in Humans and Robots, Cambridge Univ. Press, pages 39–62, 1993.Google Scholar
- 7.K. Kanatani. Geometric Computation for Machine Vision. Oxford Science Publications, 1993.Google Scholar
- 8.J. Košecká, Y. Ma, and S. Sastry. Optimization criteria, sensitivity and robustness of motion and structure estimation. In Vision Algorithms Workshop, ICCV, pages 9–16, 1999.Google Scholar
- 10.Y. Ma, J. Košecká, and S. Sastry. A mathematical theory of camera self-calibration. Electronic Research Laboratory Memorandum, UC Berkeley, UCB/ERL M98/64, October 1998.Google Scholar
- 11.Y. Ma, J. Košecká, and S. Sastry. Motion recovery from image sequences: Discrete viewpoint vs. differential viewpoint. In Proceeding of European Conference on Computer Vision, Volume II, (also Electronic Research Laboratory Memorandum M98/11, UC Berkeley), pages 337–53, 1998.Google Scholar
- 12.Y. Ma, J. Košecká, and S. Sastry. Linear differential algorithm for motion recovery: A geometric approach. Submitted to IJCV, 1999.Google Scholar
- 13.Y. Ma, J. Košecká, and S. Sastry. Optimization criteria and geometric algorithms for motion and structure estimation. submitted to IJCV, 1999.Google Scholar
- 14.S. Maybank. Theory of Reconstruction from Image Motion. Springer-Verlag, 1993.Google Scholar
- 15.J. Milnor. Morse Theory. Annals of Mathematics Studies no. 51. Princeton University Press, 1969.Google Scholar
- 16.R. M. Murray, Z. Li, and S. S. Sastry. A Mathematical Introduction to Robotic Manipulation. CRC press Inc., 1994.Google Scholar
- 17.S. Soatto and R. Brockett. Optimal and suboptimal structure from motion. Proceedings of International Conference on Computer Vision, to appear.Google Scholar
- 18.M. Spetsakis. Models of statistical visual motion estimation. CVIPG: Image Understanding, 60(3):300–312, November 1994.Google Scholar
- 19.T. Y. Tian, C. Tomasi, and D. Heeger. Comparison of approaches to egomotion computation. In CVPR, 1996.Google Scholar
- 21.J. Weng, T.S. Huang, and N. Ahuja. Motion and Structure from Image Sequences. Springer Verlag, 1993.Google Scholar
- 23.Z. Zhang. Understanding the relationship between the optimization criteria in two-view motion analysis. In Proceeding of International Conference on Computer Vision, pages 772–777, Bombay, India, 1998.Google Scholar