Share Email Print

Proceedings Paper

Neural-network-based analysis of photoelastic color using a spectral synthesizer
Author(s): Wenjun Wang; Satoru Toyooka; H. Nozawa
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Spectral distribution of photoelastic color have direct relationship with difference of principal stresses. Spectral distributions corresponding to known stress values are measured and used as learning data of unsupervised neural network. Optimal filter functions are designed by an unsupervised neural network. The learning data is also used as a scale for recognition of spectral distribution of unknown stress. Photoelastic sample under different stresses is illuminated by the light generated by a spectral synthesizer which consist of dispersing system including liquid crystal spatial light (LCSLM) modulator. By controlling the transparency of LCSLM modulator using supervised neural network, several the illuminating light are made so that their spectral distribution are same with filter functions and they are used as the light source photoelastic interferometer. From intensities recorded, spectral distributions are calculated. Stresses are estimated by comparing recorded intensities with scale made from learning data.

Paper Details

Date Published: 7 May 1999
PDF: 4 pages
Proc. SPIE 3740, Optical Engineering for Sensing and Nanotechnology (ICOSN '99), (7 May 1999); doi: 10.1117/12.347792
Show Author Affiliations
Wenjun Wang, Saitama Univ. (Japan)
Satoru Toyooka, Saitama Univ. (Japan)
H. Nozawa, Saitama Univ. (Japan)

Published in SPIE Proceedings Vol. 3740:
Optical Engineering for Sensing and Nanotechnology (ICOSN '99)
Ichirou Yamaguchi, Editor(s)

© SPIE. Terms of Use
Back to Top