FIRE operators for image processing

Fuzzy inference ruled by else-action (FIRE) operators are a class of nonlin ear operators which process image data by using fuzzy reasoning. The latest developments in the field of FIRE operators are presented in this work foc using on two very important research and application areas: nonlinear filte ring of noisy images and edge detection. First, a new family of filters for images corrupted by impulse noise is presented. Due to the adoption of pie cewise linear fuzzy sets, the proposed approach is able to combine noise ca ncellation and detail preservation. A method for automatic generation of th e fuzzy rulebase using the Genetic Algorithms is also presented. Then, a ne w class of noise-protected operators for edge detection is proposed. By sui tably choosing fuzzy sets and fuzzy aggregation mechanism, these operators are able to detect edges in images corrupted by different noise distributio ns. Many experimental results are reported showing that the proposed operat ors perform significantly better than other techniques in the literature. ( C) 1999 Elsevier Science B.V. All rights reserved.