Multigrid solution of the optical flow system using a combined diffusion‐ and curvature‐based regularizer

Abstract

Optical flow techniques are used to compute an approximate motion field in an image sequence. We apply a variational approach for the optical flow using a simple data term but introducing a combined diffusion-and curvature-based regularizer. The same data term arises in image registration problems where a deformation field between two images is computed. For optical flow problems, usually a diffusionbased regularizer should dominate, whereas for image registration a curvature-based regularizer is more appropriate. The combined regularizer enables us to handle optical flow and image registration problems with the same solver and it improves the results of each of the two regularizers used on their own. We develop a geometric multigrid method for the solution of the resulting fourth-order systems of partial differential equations associated with the variational approach for optical flow and image registration problems. The adequacy of using (collective) pointwise smoothers within the multigrid algorithm is demonstrated with the help of local Fourier analysis. Galerkin-based coarse grid operators are applied for an efficient treatment of jumping coefficients. We show some multigrid convergence rates, timings and investigate the visual quality of the approximated motion or deformation field for synthetic and real-world images.Figure 5. First line: two frames of a rotating particle sequence (size 512×512). Second line: the resulting optical flow field for = 500 at the vortex for the diffusion-based regularizer (left) and the curvature-based regularizer (right).

Improved variational image registration model and a fast algorithm for its numerical approximation

“…This approach has proved to be very useful in the understanding of MG methods when solving nonlinear problems; see for instance Refs. [20–22, 24, 29, 30, 29, 32, 51–52] for interesting examples and discussions.…”

“…For examples of successful developments in various image processing applications, refer to Refs. [20–33] where the Euler–Lagrange systems of partial differential equations (PDEs) can be of second‐ or fourth‐order. Previous work on NMG techniques for deformable image registration using free‐form or nonrigid deformations in Refs.…”

A fast algorithm for a total variation based phase demodulation model

“…It is worth to notice that although the main application of LFA is to evaluate the performance of an algorithm on linear problems with constant coefficients on an infinite grid, LFA is a tool that has been used to get some insight of discrete nonlinear operators; see the works of [17,30,[32][33][34] and the references therein. In this work, we use LFA to study the fixed point methods (8), (12) and (16) and even though they have been designed to get the solution of nonlinear problems, they can be seen as linear algorithms for u i, j k + 1 since all coefficients in Algorithm 2 are constant while iterating over q.…”

Vectorial Total Variation-Based Regularization for Variational Image Registration

“…In order to have a simple and stable numerical scheme as noted in several works in different contexts; see e.g. [25], [32] and [35, p.56-79], σ [ν] lm is approximated by…”