Cellular automata for image noise filtering

This paper presents an image noise filter based on cellular automata (CA), which can remove impulse noise from a noise corrupted image. Uniform cellular automata rules are constructed to filter impulse noise from both binary and gray scale images. Several modifications to the standard CA formulation are then applied to improve the filtering performance. For example, a random CA rule solves the noise propagation present in deterministic CA filters. A mirrored CA is used to solve the fixed boundary problem. The performance of this CA approach is compared with the classical median filter and different switching filters in terms of peak signal to noise ratio. This comparison shows that a filter based on cellular automata provides significant improvements over the standard filtering methods.