Some properties of minimizers for the Chan-Esedoglu L1TV functional

Publication date: 22 October 2007

Abstract: We present two results characterizing minimizers of the Chan-Esedoglu L1TV functional

F (u) e q u i v i n t | a b l a u | d x + l a m b d a i n t | u - f | d x

;

u, f : B b b R^{n} o B b b R

. If we restrict to

u = c h i_{S i g m a}

and

f = c h i_{O m e g a}

,

S i g m a, O m e g a i n B b b R^{n}

, the

L^{1}

TV functional reduces to

E (S i g m a) = P e r (S i g m a) + l a m b d a | S i g m a v a r t r i a n g l e O m e g a |

. We show that there is a minimizer

S i g m a

such that its boundary

p a r t i a l S i g m a

lies between the union of all balls of radius

f r a c n l a m b d a

contained in

O m e g a

and the corresponding union of

f r a c n l a m b d a

-balls in

O m e g a^{c}

. We also show that if a ball of radius

f r a c n l a m b d a + e p s i l o n

is almost contained in

O m e g a

, a slightly smaller concentric ball can be added to

S i g m a

to get another minimizer. Finally, we comment on recent results Allard has obtained on

L^{1}

TV minimizers and how these relate to our results.

This page was built for publication: Some properties of minimizers for the Chan-Esedoglu L1TV functional