A class of quasi-linear elliptic systems arising in image segmentation (Q1272325)

Let \(\Omega\) be a bounded regular domain in \(\mathbb{R}^n\) and \(\nu\) denotes the unit outer normal on the boundary \(\partial\Omega\). The authors present an existence result for weak solutions of the quasilinear elliptic system: \[ -\text{div} \bigl(a_1(z) \nabla u\bigr)+ h_1(x,z,\nabla z)u= f_1(x) \quad \text{in } \Omega,\;\partial u/ \partial v=0\text{ on } \partial\Omega, \] \[ -\text{div} \bigl(a_2(u) \nabla z\bigr)+ h_2(x,u,\nabla u)z=f_2(x) \quad\text{in }\Omega,\;\partial z/ \partial v=0\text{ on }\partial\Omega, \] where \(a_1, a_2: \mathbb{R}^n\to\mathbb{R}\) and \(h_1,h_2:\overline \Omega\times \mathbb{R}\times \mathbb{R}^n\to \mathbb{R}\) are continuous and satisfy conditions not previously considered in the literature. Their approach of linearization by iteration leads to an algorithm for a numerical approximation by finite elements. This system was motivated by the system which models image segmentation. They remark that similar results are obtainable under Dirichlet boundary conditions.