Abstract: Using the fuzzy cellular neural networks （FCNN）, the paper studies the noise suppression and character extraction in Chinese calligraphy stone inscriptions polluted by natural erosive noise. The real time stone inscription reconstruction is realized and the simulation results are given.