We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.NA

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Numerical Analysis

Title: Inhomogenous Regularization with Limited and Indirect Data

Abstract: For an ill-posed inverse problem, particularly with incomplete and limited measurement data, regularization is an essential tool for stabilizing the inverse problem. Among various forms of regularization, the lp penalty term provides a suite of regularization of various characteristics depending on the value of p. When there are no explicit features to determine p, a spatially varying inhomogeneous p can be incorporated to apply different regularization characteristics that change over the domain. This study proposes a strategy to design the exponent p when the first and second derivatives of the true signal are not available, such as in the case of indirect and limited measurement data. The proposed method extracts statistical and patch-wise information using multiple reconstructions from a single measurement, which assists in classifying each patch to predefined features with corresponding p values. We validate the robustness and effectiveness of the proposed approach through a suite of numerical tests in 1D and 2D, including a sea ice image recovery from partial Fourier measurement data. Numerical tests show that the exponent distribution is insensitive to the choice of multiple reconstructions.
Comments: 26 pages, 11 figures
Subjects: Numerical Analysis (math.NA)
MSC classes: 65F22, 65K10, 65J20
Cite as: arXiv:2108.01703 [math.NA]
  (or arXiv:2108.01703v2 [math.NA] for this version)

Submission history

From: Jihun Han [view email]
[v1] Tue, 3 Aug 2021 18:58:20 GMT (3247kb,D)
[v2] Wed, 22 Dec 2021 07:21:12 GMT (3578kb,D)

Link back to: arXiv, form interface, contact.