Investigating complexity in foggy paintings by CNN-based techniques

Aim of the paper is to deeply investigate the role and the characteristics of fog in various paintings by famous artists of different art movements, in which the presence of fog significantly affects the visual experience. To do this a new nonlinear signal processing technique, able to remove fog from color pictures exploiting optical properties of fog effect, is introduced and its implementation on a Cellular Nonlinear Network (CNN) is described. Based on the assumption that, if fog is a real element of the natural scenario, then the artist can catch its optical effect in the artwork, the proposed methodology is used to investigate whether the fog is a natural element or it has been artificially added by the artist to express its own feelings. A further analysis has been carried on to establish if the fog in some particular paintings may play the same role of noise in stochastic resonance allowing to enhance some features which cannot be distinguished in absence of fog.