A MULTIPLE-OUTPUT NONLINEAR FILTER FOR IMAGE DENOISING

F. Russo

Abstract

A multiple-output nonlinear filter for impulse noise removal from image data is presented. The proposed technique is based on the subsequent activation of two recursive filtering algorithms that operate on different subsets of input data. As a result, two pixel values are updated at each processing step producing a very effective cancellation of noise pulses. Impulse noise removal is based on rank ordered filtering. A nonlinear mechanism for error correction is also provided in order to avoid detail blur. Validation of the method is carried out by evaluating the quality of the filtered data with respect to two conflicting performance indexes: effectiveness of the noise cancellation and accuracy of detail preservation. Experimental results show that the proposed approach performs significantly better than well-known nonlinear methods in the literature including state-of-the-art operators.

Keywords

measurement of electrical quantities, noise cancellation

Download

IMEKO-WC-2000-EVL-P544.pdf

DOI

-

Back to the proceedings