Jacek MANDZIUK,1 Christophe GORECKI2E”E/sp> and Bohdan MACUKOW1

1Institute of Mathematics, Warsaw University of Technology, Plac Politechniki 1, 00-661 Warszawa, Poland, 2Laboratoire d'Optique P.M. Duffieux, (URA CNRS 214), UniversitEde Franche-ComtE 16 Route de Gray, 25030 Besanč¼Ÿn cedex, France

(Received December 20, 1995; Accepted June 3, 1996)

A new algorithm for speckle noise removal from interference patterns is proposed. The method, called cross absolute filter (CAF), is especially well suited for noisy interferograms with phase error pixels. Extensive computer simulations performed on both digitally generated fringes and experimental data show that CAF performs better than classical median-type filters, preserving more subtle details.

Key words : interferometry, speckle smoothing, non-linear filter, neural network, median filter