Noise cancellation using nonlinear fuzzy filters

Noise cancellation is a key task in the area of digital processing of measurement data. In this framework, the role of emergent techniques is rapidly growing. This paper aims at presenting the latest advances in the field of 2-D filters based on fuzzy reasoning. First, a classification of most significant approaches is proposed. Then, a collection of methods is analyzed focussing on their similarities and differences. A new filtering technique is proposed in the second part of the paper. The new filter belongs to the class of FIRE filters: it combines in the same structure rules for different noise statistics. Experimental results show that the proposed method is able to restore data corrupted by mixed Gaussian and impulse noise outperforming other techniques in the literature.