A Nonlocal Graph-PDE and Higher-Order Geometric Integration for Image Labeling

DOI10.1137/22M1496141zbMATH Open1521.39019arXiv2205.03991OpenAlexW4361290151MaRDI QIDQ6173532FDOQ6173532

Authors: Dmitrij Sitenko, Bastian Boll, Christoph Schnörr 

Publication date: 21 July 2023

Published in: SIAM Journal on Imaging Sciences (Search for Journal in Brave)

Abstract: This paper introduces a novel nonlocal partial difference equation (G-PDE) for labeling metric data on graphs. The G-PDE is derived as nonlocal reparametrization of the assignment flow approach that was introduced in extit{J.~Math.~Imaging & Vision} 58(2), 2017. Due to this parameterization, solving the G-PDE numerically is shown to be equivalent to computing the Riemannian gradient flow with respect to a nonconvex potential. We devise an entropy-regularized difference-of-convex-functions (DC) decomposition of this potential and show that the basic geometric Euler scheme for integrating the assignment flow is equivalent to solving the G-PDE by an established DC programming scheme. Moreover, the viewpoint of geometric integration reveals a basic way to exploit higher-order information of the vector field that drives the assignment flow, in order to devise a novel accelerated DC programming scheme. A detailed convergence analysis of both numerical schemes is provided and illustrated by numerical experiments.

Full work available at URL: https://arxiv.org/abs/2205.03991

Recommendations

• Image labeling by assignment
• Numerical integration of Riemannian gradient flows for image labeling
• Eikonal equation adaptation on weighted graphs: fast geometric diffusion process for local and non-local image and data processing
• Nonlocal Operators with Applications to Image Processing
• Geometric numerical integration of the assignment flow

zbMATH Keywords

geometric integrationreplicator equationinformation geometryDC programmingassignment flowsimage labelingnonlocal partial difference equation

Mathematics Subject Classification ID

Image analysis in multivariate analysis (62H35) Image processing (compression, reconstruction, etc.) in information and communication theory (94A08) Computing methodologies for image processing (68U10) Applications of difference equations (39A60) Partial difference equations (39A14) Differential geometric aspects of statistical manifolds and information geometry (53B12)

Cites Work

• ARPACK Users' Guide
• Nonlinear total variation based noise removal algorithms
• Graphical models, exponential families, and variational inference
• Title not available (Why is that?)
• Title not available (Why is that?)
• DC programming: overview.
• Deep learning
• Title not available (Why is that?)
• Fronts propagating with curvature-dependent speed: Algorithms based on Hamilton-Jacobi formulations
• A Nonlinear Conjugate Gradient Method with a Strong Global Convergence Property
• Convex Analysis
• Nonlocal discrete \(p\)-Laplacian driven image and manifold processing
• Nonlocal diffusion problems
• Nonlocal Operators with Applications to Image Processing
• Nonlocal Linear Image Regularization and Supervised Segmentation
• A generalized proximal point algorithm for certain non-convex minimization problems
• Title not available (Why is that?)
• Title not available (Why is that?)
• Analysis and approximation of nonlocal diffusion problems with volume constraints
• A nonlocal vector calculus, nonlocal volume-constrained problems, and nonlocal balance laws
• Image denoising methods. A new nonlocal principle
• Axioms and fundamental equations of image processing
• Convergence Analysis of a Proximal-Like Minimization Algorithm Using Bregman Functions
• Title not available (Why is that?)
• Title not available (Why is that?)
• The DC (Difference of convex functions) programming and DCA revisited with DC models of real world nonconvex optimization problems
• Smoothing and first order methods: a unified framework
• Motion of multiple junctions: A level set approach
• Mean curvature, threshold dynamics, and phase field theory on finite graphs
• On functions representable as a difference of convex functions
• Curvature driven interface evolution
• A combinatorial Laplacian with vertex weights
• On the \(p\)-Laplacian and \(\infty\)-Laplacian on graphs with applications in image and data processing
• Information geometry
• Image labeling by assignment
• PDE evolutions for M-smoothers in one, two, and three dimensions
• Learning adaptive regularization for image labeling using geometric assignment
• First order methods beyond convexity and Lipschitz gradient continuity with applications to quadratic inverse problems
• Continuous-domain assignment flows
• Riemannian geometry and geometric analysis
• On the geometric mechanics of assignment flows for metric data labeling
• DC programming and DCA: thirty years of developments
• Accelerating the DC algorithm for smooth functions
• Nonlocal Modeling, Analysis, and Computation
• Total Variation in Imaging
• Diffuse interface models on graphs for classification of high dimensional data
• Unsupervised assignment flow: label learning on feature manifolds by spatially regularized geometric assignment
• Learning linearized assignment flows for image labeling
• Analysis of optimization algorithms via integral quadratic constraints: nonstrongly convex problems
• Assignment flow for order-constrained OCT segmentation
• Self-assignment flows for unsupervised data labeling on graphs
• Assignment flows for data labeling on graphs: convergence and stability
• Handbook of Variational Methods for Nonlinear Geometric Data

This page was built for publication: A Nonlocal Graph-PDE and Higher-Order Geometric Integration for Image Labeling

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6173532)