APPLYING BACKPROPAGATION NEURAL NETWORKS TO FRINGE ANALYSIS

Advances in image processing and optics technology, allied to the deve lopment of algorithmic techniques such as the fast Fourier transform a nd phase stepping, have allowed automatic fringe analysis to be succes sfully applied to many problems in visual inspection and noncontact su rface measurement. However, when confronted with complicated or noisy images the algorithmic techniques tend to be less successful, implying an alternative approach may be necessary. Neural networks offer such an alternative. They have already been applied with some success to su ch conceptually similar pattern recognition problems, as the classific ation of fingerprints, the recognition of facial expressions and the i dentification of hand-written characters. Here, neural networks are ap plied to two simple fringe analysis problems. Firstly, to find the rad ius of a one-dimensional curved surface from its simulated fringe proj ection intensity distribution and, secondly, to identify four lens-sha ped objects of different radii of curvature from real fringe patterns obtained under different illumination conditions. In the first experim ent, the backpropagation and radial basis function network paradigms a re compared. In the second case, backpropagation is compared with the fuzzy-artmap paradigm. Performance criteria are the number of training data presentations, the accuracy of interpolation in the simulation e xperiment and the classification precision for the real data.