The calibration method for the Mumford-Shah functional and free-discontinuity problems (Q1871854): Difference between revisions

The Mumford-Shah functional defined as \[ F(u)=\int_{\Omega\backslash S_u}|\nabla u|^2 +\alpha{\mathcal H}^{n-1}(S_u)+\beta\int_\Omega(u-g)^2 , \] was introduced within the context of a variational approach to image segmentation problems, and can be regarded as the prototypical example of functional coupling bulk and surface contributions. In it \(\Omega\) is a bounded regular domain in \({\mathbb R}^n\), \(g: \Omega\to[0,1]\) is a given function, \(\alpha\), \(\beta\geq 0\), and \({\mathcal H}^{n-1}\) is the \((n-1)\)-dimensional Hausdorff measure. One of the most relevant features of \(F\) is a deep lack of convexity. So, even if suitable equilibrium conditions can be obtained by considering different types of infinitesimal variations, in general they do not imply minimality. In the paper a sufficient condition for minimality is proposed that closely resembles the classical principle of calibrations for minimal surfaces, and some applications of it are described. Albeit the Mumford-Shah functional cannot be included in the general theory of calibrations, some examples of calibration for \(F\) are proposed, and short and easy proofs of several minimality results are given.