New adaptive switching scheme for impulse noise removal: modelling and resolution by genetic optimisation

Abstract

Nowadays, the optimisation is one of the techniques which has proved its efficiency in many areas. In this paper, we propose a novel optimisation-based technique for impulse noise removal. Based on the classical image restoration model, we build a new objective function by introducing a new binary vector that indicates the pixels categories. We combine each pixel with the median of its neighbours in a decision rule so that one of them generates the optimal solution. The resolution of the proposed model is carried out by the genetic algorithm. Once noisy pixels are detected, a median based filter is performed only for these pixels. Experiments show that the results are satisfactory in term of both visual quality and quantitative measurement.