Neural network based automatic diffraction pattern recognition

• Autorzy: Cyran K. A. , Jaroszewicz L. R. , Niedziela T.
• Konferencja: The Second School of Photonics in Information Processing ; (7.11-11.11.2000 ; Stare Jabłonki , Poland)
• Języki publikacji: EN

Abstrakty

EN

The paper presents automatic recognition of images using the diffraction pattern sampling. This method, based on proporties of Fourier transform, utilisies special shapes of sampling element and gives the possibility to deal with data invariant with respect to typical transformation (shift, rotation and scaling) of the input images. Furthermore, if computer-generated hologram is used as feature extractor, instead of commercialy available ring-wedge detector, then the process of feature extraction can be optimised with a method proposed by the authors. The method uses rough set theory for objective function definition and stochastic evolutionary algorithms for space search. The features obtained by optimised sampling of the diffraction pattern are the input data for the semantic classifier. Since noise present in images has got typically Gaussian distribution, therefore classification should be made in the model with statistical neural network is used. The presented method is illustrated with experiments of specle pattern recognition performed with optimised and standard computer-generated holograms. The experiments confirmed good overall accuracy of the optimised system outperforming the results obtained for standard one by a factor of two to five.

Słowa kluczowe

EN

pattern recognition; computer-generated holograms; optoelectronic systems

• Wydawca: Stowarzyszenie Elektryków Polskich ; Polska Akademia Nauk ; Wojskowa Akademia Techniczna im. Jarosława Dąbrowskiego
• Czasopismo: Opto-Electronics Review
• Rocznik: 2001
• Tom: Vol. 9, No. 3
• Strony: 301--307
• Opis fizyczny: Bibliogr. 22 poz.

Twórcy

autor

Cyran K. A.

autor

Jaroszewicz L. R.

autor

Niedziela T.

• Silesian University of Technology, Institute of Computer Science, 16 Akademicka Str., 44-100 Gliwice, Poland,