Joint reconstruction in in-line holography combining parametric and non-parametric inverse approaches: application to fluid mechanics

In Vème Rencontre d’Holographie Appliquée à la Métrologie des Fluides (HOLOPHI 5)

Abstract

In-line digital holography is a simple and powerful tool to image absorbing and/or phase objects in numerous fields such as crystallography, biology or fluid mechanics. Nevertheless, this kind of interference imaging technique leads to a loss of the phase of the complex wave front on the sensor. This lack of phase information can be critical in the reconstruction process. Thus, the simplicity of the setup must be balanced by dedicated reconstruction algorithm to retrieve the object from its hologram, such as inverse approaches. In the case of simple objects for which an analytical model of propagation is known, parametric algorithms are very effective. But these approaches fail at reconstructing more complex objects, where non-parametric solutions must be involved. This may lead to a loss in precision or specificity. In this work we propose a new approach combining these two methods to take benefits from their own advantages. The object to reconstruction is split in two subparts. A part is described by a parametric model. The other part of the object is simulated via a nonparametric model. These two parts which interfere are jointly considered in the reconstruction algorithm by alternating parametric and non-parametric procedures. We apply this new technique to evaporating droplets where the high contrast fringes produced by the droplets tend to mask the fringes produced by the plume. With our method, both the droplet and the plume are jointly reconstructed.