Image Processing by a Discrete Reaction-Diffusion System

We propose a new concept of image processing by use of self-organization mechanisms appeared in a discrete non-linear system. Edge detection, image segmentation, noise reduction and contrast enhancement can be achieved by use of a discrete reaction-diffusion model (Fitz-Hugh & Nagumo model) under the condition of Turing instability. Compared with the conventional method, the proposed one indicates a higher performance in processing for noisy image. Especially, by adding a suitable level of noise, the model works as contrast enhancement. It is confirmed that the non-linear effect of stochastic resonance (SR) brings a good performance in image processing