-
The Proxy Step-size Technique for Regularized Optimization on the Sphere Manifold
Authors:
Fang Bai,
Adrien Bartoli
Abstract:
We give an effective solution to the regularized optimization problem $g (\boldsymbol{x}) + h (\boldsymbol{x})$, where $\boldsymbol{x}$ is constrained on the unit sphere $\Vert \boldsymbol{x} \Vert_2 = 1$. Here $g (\cdot)$ is a smooth cost with Lipschitz continuous gradient within the unit ball $\{\boldsymbol{x} : \Vert \boldsymbol{x} \Vert_2 \le 1 \}$ whereas $h (\cdot)$ is typically non-smooth b…
▽ More
We give an effective solution to the regularized optimization problem $g (\boldsymbol{x}) + h (\boldsymbol{x})$, where $\boldsymbol{x}$ is constrained on the unit sphere $\Vert \boldsymbol{x} \Vert_2 = 1$. Here $g (\cdot)$ is a smooth cost with Lipschitz continuous gradient within the unit ball $\{\boldsymbol{x} : \Vert \boldsymbol{x} \Vert_2 \le 1 \}$ whereas $h (\cdot)$ is typically non-smooth but convex and absolutely homogeneous, \textit{e.g.,}~norm regularizers and their combinations. Our solution is based on the Riemannian proximal gradient, using an idea we call \textit{proxy step-size} -- a scalar variable which we prove is monotone with respect to the actual step-size within an interval. The proxy step-size exists ubiquitously for convex and absolutely homogeneous $h(\cdot)$, and decides the actual step-size and the tangent update in closed-form, thus the complete proximal gradient iteration. Based on these insights, we design a Riemannian proximal gradient method using the proxy step-size. We prove that our method converges to a critical point, guided by a line-search technique based on the $g(\cdot)$ cost only. The proposed method can be implemented in a couple of lines of code. We show its usefulness by applying nuclear norm, $\ell_1$ norm, and nuclear-spectral norm regularization to three classical computer vision problems. The improvements are consistent and backed by numerical experiments.
△ Less
Submitted 5 September, 2022;
originally announced September 2022.
-
An Optimal Triangle Projector with Prescribed Area and Orientation, Application to Position-Based Dynamics
Authors:
Carlos Arango Duque,
Adrien Bartoli
Abstract:
The vast majority of mesh-based modelling applications iteratively transform the mesh vertices under prescribed geometric conditions. This occurs in particular in methods cycling through the constraint set such as Position-Based Dynamics (PBD). A common case is the approximate local area preservation of triangular 2D meshes under external editing constraints. At the constraint level, this yields t…
▽ More
The vast majority of mesh-based modelling applications iteratively transform the mesh vertices under prescribed geometric conditions. This occurs in particular in methods cycling through the constraint set such as Position-Based Dynamics (PBD). A common case is the approximate local area preservation of triangular 2D meshes under external editing constraints. At the constraint level, this yields the nonconvex optimal triangle projection under prescribed area problem, for which there does not currently exist a direct solution method. In current PBD implementations, the area preservation constraint is linearised. The solution comes out through the iterations, without a guarantee of optimality, and the process may fail for degenerate inputs where the vertices are colinear or colocated. We propose a closed-form solution method and its numerically robust algebraic implementation. Our method handles degenerate inputs through a two-case analysis of the problem's generic ambiguities. We show in a series of experiments in area-based 2D mesh editing that using optimal projection in place of area constraint linearisation in PBD speeds up and stabilises convergence.
△ Less
Submitted 24 November, 2020;
originally announced November 2020.
-
Solutions of Quadratic First-Order ODEs applied to Computer Vision Problems
Authors:
David Casillas-Perez,
Daniel Pizarro,
Manuel Mazo,
Adrien Bartoli
Abstract:
This article is a study about the existence and the uniqueness of solutions of a specific quadratic first-order ODE that frequently appears in multiple reconstruction problems. It is called the \emph{planar-perspective equation} due to the duality with the geometric problem of reconstruction of planar-perspective curves from their modulus. Solutions of the \emph{planar-perspective equation} are re…
▽ More
This article is a study about the existence and the uniqueness of solutions of a specific quadratic first-order ODE that frequently appears in multiple reconstruction problems. It is called the \emph{planar-perspective equation} due to the duality with the geometric problem of reconstruction of planar-perspective curves from their modulus. Solutions of the \emph{planar-perspective equation} are related with planar curves parametrized with perspective parametrization due to this geometric interpretation. The article proves the existence of only two local solutions to the \emph{initial value problem} with \emph{regular initial conditions} and a maximum of two analytic solutions with \emph{critical initial conditions}. The article also gives theorems to extend the local definition domain where the existence of both solutions are guaranteed. It introduces the \emph{maximal depth function} as a function that upper-bound all possible solutions of the \emph{planar-perspective equation} and contains all its possible \emph{critical points}. Finally, the article describes the \emph{maximal-depth solution problem} that consists of finding the solution of the referred equation that has maximum the depth and proves its uniqueness. It is an important problem as it does not need initial conditions to obtain the unique solution and its the frequent solution that practical algorithms of the state-of-the-art give.
△ Less
Submitted 27 June, 2018; v1 submitted 11 October, 2017;
originally announced October 2017.
-
Rank-constrained fundamental matrix estimation by polynomial global optimization versus the eight-point algorithm
Authors:
Florian Bugarin,
Adrien Bartoli,
Didier Henrion,
Jean-Bernard Lasserre,
Jean-José Orteu,
Thierry Sentenac
Abstract:
The fundamental matrix can be estimated from point matches. The current gold standard is to bootstrap the eight-point algorithm and two-view projective bundle adjustment. The eight-point algorithm first computes a simple linear least squares solution by minimizing an algebraic cost and then projects the result to the closest rank-deficient matrix. We propose a single-step method that solves both s…
▽ More
The fundamental matrix can be estimated from point matches. The current gold standard is to bootstrap the eight-point algorithm and two-view projective bundle adjustment. The eight-point algorithm first computes a simple linear least squares solution by minimizing an algebraic cost and then projects the result to the closest rank-deficient matrix. We propose a single-step method that solves both steps of the eight-point algorithm. Using recent results from polynomial global optimization, our method finds the rank-deficient matrix that exactly minimizes the algebraic cost. In this special case, the optimization method is reduced to the resolution of very short sequences of convex linear problems which are computationally efficient and numerically stable. The current gold standard is known to be extremely effective but is nonetheless outperformed by our rank-constrained method for bootstrapping bundle adjustment. This is here demonstrated on simulated and standard real datasets. With our initialization, bundle adjustment consistently finds a better local minimum (achieves a lower reprojection error) and takes less iterations to converge.
△ Less
Submitted 19 March, 2014;
originally announced March 2014.
