论文信息 - Pattern recognition neural-net by spatial mapping of biology visual field

Pattern recognition neural-net by spatial mapping of biology visual field

The method of spatial mapping in biology vision field is applied to artificial neural networks for pattern recognition. By the coordinate transform that is called the complex-logarithm mapping and Fourier transform, the input images are transformed into scale- rotation- and shift- invariant patterns, and then fed into a multilayer neural network for learning and recognition. The results of computer simulation and an optical experimental system are described.

Xin Lin | Masahiko Mori

[1] M. Mori,et al. Improvement of the generalization capability for a pattern-recognition neural network that uses a Gaussian-synapse neuron model. , 2000, Applied optics.

[2] D. Casasent,et al. SCALE INVARIANT OPTICAL CORRELATION USING MELLIN TRANSFORMS , 1976 .

[3] Shigeo Abe,et al. Dependency of Generalization Capability for a Multi-Layered Neural Network on its Number of Hidden Units , 1993 .

[4] M Fang,et al. Class of transforms invariant under shift, rotation, and scaling. , 1990, Applied optics.

[5] J. O'Regan,et al. Some results on translation invariance in the human visual system. , 1990, Spatial vision.

[6] T Yatagai,et al. Optical learning neural network with a pockels readout optical modulator. , 1998, Applied optics.

[7] J Shamir,et al. Target location measurement by optical correlators: a performance criterion. , 1992, Applied optics.

[8] P Cavanagh,et al. Size and Position Invariance in the Visual System , 1978, Perception.