Fuzzy Logic based Improvement in Discrete Wavelet Transform Edge
Representation in Images
Abstract
We introduce a rule base fuzzy technique on decomposed wavelet
coefficients, to improve the wavelet edge representation. Our algorithm
mitigates ‘incorrect’ responses, due primarily to the symmetries of
directional derivative filters. Since the Discrete Wavelet Transform
(DWT) coefficients are revealed from two dimensional symmetric filter
banks and undermine some gradient information. These wavelet
coefficients are prearranged into ‘if-then’ rule structure of a fuzzy
inference system, to improve the wavelet edge representation.