Interferogram Analysis using Active Instance-based Learning

O. Fuentes and T. Solorio (Mexico)


active learning, instance-based learning, locally weighted regression


In this paper we present an efficient solution, based on machine learning, to the problem of obtaining the phase of a set of interferograms. The algorithm learns the function from an interferogram, given as a gray scale image, to a vector of aberration coefficients that form the Seidel aberration representation of the in terferogram. The algorithm uses principal component analysis to reduce the dimensionality of the task and applies a modified version to the locally-weighted re gression algorithm to find aberration coefficients. The method is faster than others previously presented in the literature, and is also noise-insensitive. We tested our algorithm using a large set of simulated interfero grams, and our experiments show very accurate results using both noiseless and noisy interferograms.

Important Links:

Go Back