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

Download:

Current browse context:

cs.CV
< prev | next >
new | recent | 2106

Change to browse by:

cs
cs.SC
eess
eess.IV

References & Citations

DBLP - CS Bibliography

listing | bibtex

Bookmark

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

Computer Science > Computer Vision and Pattern Recognition

Title: Design of Low-Artifact Interpolation Kernels by Means of Computer Algebra

Authors: Peter Karpov
(Submitted on 8 Jun 2021)
Abstract: We present a number of new piecewise-polynomial kernels for image interpolation. The kernels are constructed by optimizing a measure of interpolation quality based on the magnitude of anisotropic artifacts. The kernel design process is performed symbolically using Mathematica computer algebra system. Experimental evaluation involving 14 image quality assessment methods demonstrates that our results compare favorably with the existing linear interpolators.
Comments: 22 pages, 6 figures
Subjects: Computer Vision and Pattern Recognition (cs.CV); Symbolic Computation (cs.SC); Image and Video Processing (eess.IV)
Cite as: arXiv:2106.04104 [cs.CV]
  (or arXiv:2106.04104v1 [cs.CV] for this version)

Submission history

From: Peter Karpov [view email]
[v1] Tue, 8 Jun 2021 05:06:51 GMT (5729kb,D)
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Link back to: arXiv, form interface, contact.