Covariance estimation for minimal geometry solvers via scaled unscented transformation

Abstract Covariance is a well-established characterisation of the output uncertainty for estimators dealing with noisy data. It is conventionally estimated via first-order forward propagation (FOP) of input covariance. However, since FOP employs a local linear approximation of the estimator, its reliability is compromised in the case of nonlinear transformations. An alternative method, scaled unscented transformation (SUT) is known to cope with such cases better. However, despite the nonlinear nature of many vision problems, its adoption remains limited. This paper investigates the application of SUT on common minimal geometry solvers, a class of algorithms at the core of many applications ranging from image stitching to film production and robot navigation. The contributions include an experimental comparison of SUT against FOP on synthetic and real data, and practical suggestions for adapting the original SUT to the geometry solvers. The experiments demonstrate the superiority of SUT to FOP as a covariance estimator, over a range of scene types and noise levels, on synthetic and real data.

[1]  Bernhard P. Wrobel,et al.  Multiple View Geometry in Computer Vision , 2001 .

[2]  Olivier D. Faugeras,et al.  Characterizing the Uncertainty of the Fundamental Matrix , 1997, Comput. Vis. Image Underst..

[3]  Robert M. Haralick,et al.  Propagating covariance in computer vision , 1994, Proceedings of 12th International Conference on Pattern Recognition.

[4]  Manolis I. A. Lourakis,et al.  Estimating the Jacobian of the Singular Value Decomposition: Theory and Applications , 2000, ECCV.

[5]  Klas Josephson,et al.  Pose estimation with radial distortion and unknown focal length , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[6]  Joachim Denzler,et al.  Global Uncertainty-based Selection of Relative Poses for Multi Camera Calibration , 2008, BMVC.

[7]  Robert M. Haralick.  Propagating covariance in computer vision , 1994, Proceedings of 12th International Conference on Pattern Recognition.

[8]  Ian D. Reid,et al.  A plane measuring device , 1999, Image Vis. Comput..

[9]  S. Roberts,et al.  Bayesian methods for autoregressive models , 2000, Neural Networks for Signal Processing X. Proceedings of the 2000 IEEE Signal Processing Society Workshop (Cat. No.00TH8501).

[10]  Olivier Stasse,et al.  MonoSLAM: Real-Time Single Camera SLAM , 2007, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  J. Clarke.  Modelling uncertainty: A primer , 1998 .

[12]  Nando de Freitas,et al.  The Unscented Particle Filter , 2000, NIPS.

[13]  Adrien Bartoli,et al.  Nonlinear estimation of the fundamental matrix with minimal parameters , 2004, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[14]  Berthold K. P. Horn,et al.  Closed-form solution of absolute orientation using unit quaternions , 1987 .

[15]  Thorsten Thormählen,et al.  Keyframe Selection for Camera Motion and Structure Estimation from Multiple Views , 2004, ECCV.

[16]  Robert C. Bolles,et al.  Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography , 1981, CACM.

[17]  Jean-Yves Guillemaut,et al.  Calibration of Nodal and Free-Moving Cameras in Dynamic Scenes for Post-Production , 2011, 2011 International Conference on 3D Imaging, Modeling, Processing, Visualization and Transmission.

[18]  Robert M. Haralick,et al.  Analysis and solutions of the three point perspective pose estimation problem , 1991, Proceedings. 1991 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[19]  Kenichi Kanatani,et al.  Analysis of 3-D Rotation Fitting , 1994, IEEE Trans. Pattern Anal. Mach. Intell..

[20]  Marie-Odile Berger,et al.  Computing the Uncertainty of the 8 point Algorithm for Fundamental Matrix Estimation , 2008, BMVC.

[21]  Michael Werman,et al.  How to put probabilities on homographies , 2005, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[22]  David Nistér,et al.  An efficient solution to the five-point relative pose problem , 2003, 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings..

[23]  David W. Murray,et al.  A Square Root Unscented Kalman Filter for visual monoSLAM , 2008, 2008 IEEE International Conference on Robotics and Automation.

[24]  F.,et al.  Quaternion Averaging , 2007 .

[25]  Jeffrey K. Uhlmann,et al.  Unscented filtering and nonlinear estimation , 2004, Proceedings of the IEEE.

[26]  E. Kraft,et al.  A quaternion-based unscented Kalman filter for orientation tracking , 2003, Sixth International Conference of Information Fusion, 2003. Proceedings of the.

[27]  Wolfgang Förstner,et al.  Minimal Representations for Uncertainty and Estimation in Projective Spaces , 2010, ACCV.

[28]  Jan-Michael Frahm,et al.  Exploiting uncertainty in random sample consensus , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[29]  Stephan Heuel,et al.  Uncertain Projective Geometry: Statistical Reasoning for Polyhedral Object Reconstruction , 2004, Lecture Notes in Computer Science.

[30]  Zuzana Kukelova,et al.  Polynomial Eigenvalue Solutions to the 5-pt and 6-pt Relative Pose Problems , 2008, BMVC.

[31]  Eduardo Mario Nebot,et al.  Consistency of the EKF-SLAM Algorithm , 2006, 2006 IEEE/RSJ International Conference on Intelligent Robots and Systems.