Bayesian Optical Flow with Uncertainty Quantification

Optical flow refers to the visual motion observed between two consecutive images. Since the degree of freedom is typically much larger than the constraints imposed by the image observations, the straightforward formulation of optical flow as an inverse problem is ill-posed. Standard approaches to determine optical flow rely on formulating and solving an optimization problem that contains both a data fidelity term and a regularization term, the latter effectively resolves the otherwise ill-posedness of the inverse problem. In this work, we depart from the deterministic formalism, and instead treat optical flow as a statistical inverse problem. We discuss how a classical optical flow solution can be interpreted as a point estimate in this more general framework. The statistical approach, whose "solution" is a distribution of flow fields, which we refer to as Bayesian optical flow, allows not only "point" estimates (e.g., the computation of average flow field), but also statistical estimates (e.g., quantification of uncertainty) that are beyond any standard method for optical flow. As application, we benchmark Bayesian optical flow together with uncertainty quantification using several types of prescribed ground-truth flow fields and images.

[1]  Albert Akhriev,et al.  Dynamic cloud motion forecasting from satellite images , 2017, 2017 IEEE 56th Annual Conference on Decision and Control (CDC).

[2]  Jean-Michel Morel,et al.  A Nonlocal Bayesian Image Denoising Algorithm , 2013, SIAM J. Imaging Sci..

[3]  Joachim Weickert,et al.  Variational Optic Flow Computation with a Spatio-Temporal Smoothness Constraint , 2001, Journal of Mathematical Imaging and Vision.

[4]  A. Tikhonov,et al.  Numerical Methods for the Solution of Ill-Posed Problems , 1995 .

[5]  Johnathan M. Bardsley,et al.  MCMC-Based Image Reconstruction with Uncertainty Quantification , 2012, SIAM J. Sci. Comput..

[6]  L. Cavalier Nonparametric statistical inverse problems , 2008 .

[7]  E. Bollt,et al.  A Multi-time Step Method to Compute Optical Flow with Scientific Priors for Observations of a Fluidic System , 2014 .

[8]  Isabelle Herlin,et al.  Solving ill-posed Image Processing problems using Data Assimilation , 2011, Numerical Algorithms.

[9]  C. Vogel Computational Methods for Inverse Problems , 1987 .

[10]  Kahlouche Souhila,et al.  Optical Flow Based Robot Obstacle Avoidance , 2007 .

[11]  Isabelle Herlin,et al.  Divergence-Free Motion Estimation , 2012, ECCV.

[12]  P. Hansen Rank-Deficient and Discrete Ill-Posed Problems: Numerical Aspects of Linear Inversion , 1987 .

[13]  Berthold K. P. Horn,et al.  Determining Optical Flow , 1981, Other Conferences.

[14]  Isabelle Herlin,et al.  Optical Flow and Phase Portrait Methods for Environmental Satellite Image Sequences , 1996, ECCV.

[15]  E. Bollt,et al.  A Lagged Diffusivity Method for Computing Total Variation Regularized Fluid Flow , 2013 .

[16]  Damien Garcia,et al.  A Proof of Convergence of the Horn-Schunck Optical Flow Algorithm in Arbitrary Dimension , 2014, SIAM J. Imaging Sci..

[17]  Erik Bollt,et al.  Data fusion reconstruction of spatially embedded complex networks , 2017, J. Complex Networks.

[18]  David J. Fleet,et al.  Performance of optical flow techniques , 1992, Proceedings 1992 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[19]  Johnathan M. Bardsley,et al.  Gaussian Markov random field priors for inverse problems , 2013 .

[20]  Dana H. Ballard,et al.  Computer Vision , 1982 .

[21]  E. Bollt,et al.  A framework for estimating potential fluid flow from digital imagery. , 2013, Chaos.

[22]  D. Higdon Space and Space-Time Modeling using Process Convolutions , 2002 .

[23]  A. N. Tikhonov,et al.  REGULARIZATION OF INCORRECTLY POSED PROBLEMS , 1963 .

[24]  Giannis K. Chantas,et al.  Variational-Bayes Optical Flow , 2014, Journal of Mathematical Imaging and Vision.

[25]  Michael J. Black,et al.  Secrets of optical flow estimation and their principles , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[26]  F. O’Sullivan A Statistical Perspective on Ill-posed Inverse Problems , 1986 .

[27]  Aaron Luttman,et al.  A Stream Function Approach to Optical Flow with Applications to Fluid Transport Dynamics , 2011 .

[28]  D. M. Titterington,et al.  A Study of Methods of Choosing the Smoothing Parameter in Image Restoration by Regularization , 1991, IEEE Trans. Pattern Anal. Mach. Intell..

[29]  Albert Akhriev,et al.  Where computer vision can aid physics: dynamic cloud motion forecasting from satellite images , 2017, ArXiv.

[30]  Matthias W. Seeger,et al.  Large Scale Bayesian Inference and Experimental Design for Sparse Linear Models , 2011, SIAM J. Imaging Sci..

[31]  David J. Fleet,et al.  Performance of optical flow techniques , 1994, International Journal of Computer Vision.

[32]  Rachid Deriche,et al.  Computing Optical Flow via Variational Techniques , 1999, SIAM J. Appl. Math..

[33]  Mlnoru Slotani Tolerance regions for a multivariate normal population , 1964 .