Interference fringes analysis based on recurrence non-linear filtering methodology

The interferometry technique is widely used in dimensional metrology, optical testing and modern precision technologies. The intensity of interference fringes depends on phase, frequency or visibility maximum position non-linearly, but namely the parameters pointed contain the useful information about the object to be tested. Well known methods based on the recovery of analytical signal parameters and the phase shifting interferometry technique have not enough noise-immunity and require to register the whole data series before parameter calculation. We propose a new approach to interference fringes analysis. Interferometric data are assumed to be a series of samples of the interferometric signal defined by known model. The signal value from the previous step to the next step is predicted. The prediction error is used for signal parameters correction. Recurrence non-linear processing generally gives the optimal dynamic assessment of phase, frequency and visibility maximum position. Stability and accuracy of recurrence non-linear filtering of interference fringes were investigated. Optimal nonlinear algorithms were developed with application to conventional one-wavelength interferometry, to two-wavelength and low coherent interferometers.