Total variation regularization superior to classical quadratic choice

Mathematics Asked by Pazu on December 9, 2020

Where can I find literature on why the choice of regularization $$int_{Omega} |nabla u| ,{rm d} x,{rm d}y$$ is more effective than the classical choice $$int_{Omega} |nabla u|^2 ,{rm d} x,{rm d}y$$ when it comes to inverse problems occuring in the context of images?

image processing inverse problems reference request regularization total variation

One Answer

It has been discussed it pretty much every numerical inverse problems talk that uses TV regularization. Presumably following the references from any of the associated papers, or any paper in image processing that uses the technique, would lead you to good sources.

And indeed: The first one I found cites Rudin-Osher-Fatemi (also cited in the comments by Eric Towers)

The main heuristic seems to be: If you want smooth results, you should use $L^2$ regularization. If you want sharp edges, and artifical staircasing artefacts are not a huge issue, you should consider TV.

Correct answer by Tommi on December 9, 2020

Total variation regularization superior to classical quadratic choice

One Answer

Add your own answers!

Related Questions

(proposed) elegant solution to IMO 2003 P1

Reverse order of polynomial coefficients of type $left(r-xright)^n$

Should one include already cemented proofs of related principles in one’s paper?

Find the smallest $k$ such that $kx bmod n in [a,b]$

The Cauchy-Crofton formula on a plane

Solving a limit for capacity of a transmission system

Galois group Abstract algebra

Does there always exists coefficients $c,dinmathbb{R}$ s.t. $ax^3+bx^2+cx+d$ has three different real roots?

If there is a “worldly ordinal,” then must there be a worldly cardinal?

Cover number and matching number in hypergraphs.

$displaystyleint_C (e^x+cos(x)+2y),dx+(2x-frac{y^2}{3}),dy$ in an ellipse

Show that $h_n(x)=x^{1+frac{1}{2n-1}}$ converges uniformly on $[-1, 1]$.

What does the functional monotone class theorem say and how does it relate to the other monotone class theorem?

Symmetric matrix and Hermitian matrix, unitarily diagonalizable

What remarks about the tails can we infer from the density function of $X^n$ if $X$ is standard normally distributed

direct conversion from az/el to ecliptic coordinates

A proof problem in mathematical statistics

How to find the z component of the parameterization of an ellipse that is the intersection of a vertical cylinder and a plane

using Lagrange multipliers to determine shortest distance between a point and straight line

Proof of Lemma 5.1.5.3 in Jacob Lurie’s HTT.

Ask a Question