Intelligent digital filters with application to salt and pepper noise reduction in MR brain images

This paper introduces intelligent digital filters and demonstrates its operations using a fuzzy-based intelligent digital filter with an application to salt and pepper noise reduction in magnetic resonance (MR) brain images. The fuzzy-based intelligent digital filter adopts an adaptive window size, a salt and pepper noise detector, and a novel membership function. It employs an information-to-coefficient processor for computing input-dependent filter coefficients and a fuzzy filter for time-varying digital filtering. Simulation results indicate that the filter is effective for reducing low to high density salt and pepper noise in MR brain images. Performance comparisons to the iterative median filter and the non-local means filter are included.