A neural network model for solving nonlinear optimization problems with real-time applications

Abstract

A new neural network model is proposed for solving nonlinear optimization problems with a general form of linear constraints. Linear constraints, which may include equality, inequality and bound constraints, are considered to cover the need for engineering applications. By employing this new model in image fusion algorithm, an optimal fusion vector is exploited to enhance the quality of fused images efficiently. The stability and convergence analysis of the novel model are proved in details. The simulation examples are used to demonstrate the validity of the proposed model. © 2009 Springer Berlin Heidelberg.