Regensburg 2004 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe

MA: Magnetismus

MA 7: Magnetische Abbildungsverfahren

MA 7.2: Vortrag

Montag, 8. März 2004, 15:30–15:45, H22

Noise and artifact suppression for near-field magneto-optical microscopy through improved data preprocessing and piloted wavelet analysis — •Fabian Kiendl and Gernot Güntherodt — II. Physikalisches Institut der RWTH Aachen, D-52056 Aachen, Germany

The extraction of information about a sample from the raw data acquired by our magneto-optical SNOM set-up [1,2] needs to be improved. We propose a refined pre-processing of raw data into a raw image and a wavelet de-noising that is tailored to SNOM images.

We make the pre-processing more resilient to outliers, base it additionally on previously unused portions of the raw data, and refine the suppression of low-frequency noise. Image features are pronounced much more clearly in the resulting raw image, and the previously present stripe-like artifacts are removed. To tackle high-frequency noise, we adjust the parameters of wavelet de-noising [3,4] to a given noisy SNOM image by de-noising a superposition of this SNOM image with a known pilot image. Additionally, we overcome the tendency of wavelet de-noising to produce directional artifacts.

We find that images are de-noised well enough for our recently proposed deconvolution method [5] to further enhance them. Yet we find de-noising not to be restricted to SNOM images.

[1] G. Eggers, PhD thesis, Aachen University, ISBN 3-89825-065-2 (1999)

[2] A. Rosenberger, PhD thesis, Aachen University (2000)

[3] F. Kiendl, G. Güntherodt, 269th WE-Heraeus seminar (2002)

[4] W. Bäni, Wavelets, ISBN 3-486-25427-8, Oldenbourg (2002)

[5] F. Kiendl, G. Güntherodt, Verh. DPG (VI) 37, 1/215 (2002)