The application of parametric estimation techniques to wavelength scanning interferometry

The growth in the manufacture of complex surfaces means that surface metrology has become an essential part of the manufacturing process in many industries. Wavelength scanning interferometry (WSI) is a technique that enables fast surface topography measurement. Since WSI can be implemented without the measurand or apparatus being mechanically moved, the measurement times are substantially less than other topography techniques that require mechanical movement. A variety of algorithms can be used to evaluate the spectral interferograms generated from WSI, and the choice of algorithm affects the axial (height) resolution and measurement range that can be attained. At present Fourier transform based methods are the most commonly used, but these can have a number of limitations, including problems which derive from short data lengths.

In order to extend the measurement range of the WSI, frequency estimation techniques based on an auto-regressive model are introduced. In particular, the Burg method is suggested as an alternative to current algorithms and a comparison is made to Fourier-transform methods using both simulated data and experimental data of a step-height sample measured using WSI.