Nonlinear tomographic reconstruction via nonsmooth optimization
This interesting paper deals with the subject of computed tomography scanning. What is a computed tomography (CT) scan? Essentially a CT scan is a procedure in image processing where a motorized X-ray source rotates rapidly around an object of interest, emitting narrow beams of X-rays at the object at varying angles. Detectors are then put on opposite sides of the X-ray and they measure the transmitted radiation. Then the measurements (at different anges) are combined to produce various images of the object for example cross-sectional images. In this paper, the authors study the case when measurements are produced by a linear transformation of an unknown signal followed by an exponential nonlinear map. The authors illustrate that a initialized subgradient method applied to a certain natural nonsmooth, nonconvex loss function produces iterates that converge to the unknown signal of interest at a geometric rate. \N\NThe paper is well wtitten with a good set of references.
- Approximate solution of systems of linear equations†
- Generalized Gradients and Applications
- scientific article; zbMATH DE number 3008728 (Why is no real title available?)
- Implicit regularization in nonconvex statistical estimation: gradient descent converges linearly for phase retrieval, matrix completion, and blind deconvolution
- Low-rank matrix recovery with composite optimization: good conditioning and rapid convergence
- Minimization of unsmooth functionals
- Nonconvex Robust Low-Rank Matrix Recovery
- Nonlinear total variation based noise removal algorithms
- On convergence rates of subgradient optimization methods
- Sharp global convergence guarantees for iterative nonconvex optimization with random data
- Stochastic (Approximate) Proximal Point Methods: Convergence, Optimality, and Adaptivity
- Stochastic algorithms with geometric step decay converge linearly on sharp functions
- Structured Signal Recovery From Quadratic Measurements: Breaking Sample Complexity Barriers via Nonconvex Optimization
- Subgradient methods for sharp weakly convex functions
- The landscape of empirical risk for nonconvex losses
- The mathematics of computerized tomography
- Why do commercial CT scanners still employ traditional, filtered back-projection for image reconstruction?
- Über die Bestimmung von Funktionen durch ihre Integralwerte längs gewisser Mannigfaltigkeiten.
This page was built for publication: Nonlinear tomographic reconstruction via nonsmooth optimization
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6956546)
