Find in Library

We present an analysis of the recognition performance of 3-D objects reconstructed from digital holograms recorded under photon counting conditions. The digital holograms are computed by applying four-step phase-shifting techniques to interferograms recorded with weak coherent light. Recognition capability is analyzed as a function of the total number of photons by using a maximum-likelihood approach adapted to one-class classification problems. The likelihood is modeled assuming a Gaussian distribution, whose centroid corresponds to the highest value in a mixture of two Gaussian values. The recognition capability is studied both in terms of the axial distance and the lateral position of the reconstructed 3-D object.