We propose a parametric method to estimate geometrical properties of a population of spheroid-like particles from 2D projection images. The method consists in, first, detecting the projection of the particles in the images, and then estimating the parameters of the supposed probability laws of the spheroids semi-axes using a Bayesian framework. Moreover, a new estimator of the Sauter mean diameter to assess the efficiency of two-phase flow processes, in the case of spheroid-like particle system, is proposed. Still in view to its practical use for the characterization of two-phase flows, the whole methodology is applied to a typical bubbly flow.