Local signal processing to evaluate resolution in SNOM images, using 1D wavelets

The near-field optical images are known to be inhomogeneous due on the one hand to the piezo decay and on the other hand to the scanning process. Both can induce a variation of the tip-sample distance between the beginning and the end of the scanning in constant height mode and steps between consecutive scanning lines. Moreover, the tip often rubs on the sample and induces artifactual, local high variations of the recorded signal. Therefore, a local characterization of spectral information in the signal could be useful in defining the separation power of the setup. This paper proposes a convenient method to evaluate this local spectral information in terms of local characterization of spatial frequency power in the recorded signal. The location of the spatial frequencies is achieved by using the discrete wavelet transform method. (C) 2000 Elsevier Science B.V. All rights reserved.