AI & ML
Total Generalized Variation for Inverse Problems Involving Piecewise Linear Finite Elements
arXiv cs.NA0 views
arXiv:2609.12778v2 Announce Type: replace Abstract: Higher-order regularization has shown improved results over first-order methods such as total variation. Piecewise linear discretizations are required in applications such as electrocardiographic imaging, where the forward operator is only defined for weakly differentiable functions. In this paper, convergence of a piecewise linear finite element discretization is shown for the non-symmetric second-order total generalized variation functional in a denoising setting. Both the function and the auxiliary variable are discretized with piecewise linear finite elements. Additionally, this convergence result is extended to a spatiotemporal setting after the spatiotemporal function space is comprehensively introduced. In both cases, discrete minimizers converge to the continuous minimizer with rate $h^{1/4}$ in the mesh size $h$, matching the best known rate for finite element discretizations of total variation in the same general setting. Numerical experiments on regular and unstructured meshes are consistent with the theoretical convergence rates and indicate faster convergence in practice, with observed rates of $h^{1/2}$ in the spatial and $h$ in the spatiotemporal setting relative to a high-resolution reference solution. For image denoising and inpainting on a standard benchmark dataset, reconstruction quality is comparable to related piecewise constant and piecewise linear methods, while using half the degrees of freedom of the piecewise constant discretization. The spatiotemporal case is discussed using the example of the inverse problem in electrocardiographic imaging. On three synthetic two-dimensional test cases, the proposed functional outperforms Tikhonov and spatiotemporal total variation regularization, reducing the relative error by 12.5 to 43.1% over first-order Tikhonov regularization.
Source:https://arxiv.org/abs/2609.12778PreviousRandomized Block Davidson Eigensolvers for Plane-Wave Density-Functional TheoryNextFast Mixed-Precision Real Evaluation
Related posts
Best time trackers for developers, for code and for AI
0dev.to

Serving Qwen3.5-0.8B with a 0.016% weight edit: measure the loops, pin the sampling on the server
0dev.to
Design Tokens for AI: Extract a Website's Real Colours, Fonts, Shadows and Icons So an LLM Builds On Brand
0dev.to
Trying to get our Arms around AI efficiency
0Stack Overflow Blog
Why an Embedded Game Fails on Some Phones, and How to Debug It
0dev.to
Microsoft 365 vs Office 2024: The Break-Even Math
0dev.to