Retrieval of particle parameters with the neural network using multi-angle
light-scattering data
V. Berdnik1, O. Gritsai2 and V. Loiko*,2
- 1 Techniques of Hydrogeological Investigation S.A., Kazan, 420015 Russia.
- 2 Institute of Physics of the NASB, Nezavisimosti ave. 68, Minsk, 220072 Belarus.
A method to retrieve radius and refractive index of spherical homogeneous non-absorbing particles by multi-angle scattering is proposed. It is based on formation of the noise resistant functionals of intensity, which are invariant relative to linear homogeneous transformations of an intensity-based signal, and approximation of the retrieved parameters' dependence on the functionals by a feed-forward neural network. The method allows reducing retrieval errors in the region of small sizes and refractive indices of particles close to unity.
- * Corresponding author