Multistable cellular neural networks and their application to image decomposition
José Antonio Medina Hernández, Felipe Gómez Castañeda, José Antonio Moreno Cadenas · 2009
Traditional cellular neural networks are arrays of coupled processing cells, where every uncoupled cell has two stationary stable states. In this paper an oscillatory function is used to define a cellular neural network whose uncoupled cells have a number of stable stationary states larger than two, so the output image has three or more grey levels. The proposed dynamics is applied to the decomposition of images with multiple gray levels.