Inversion of spectroscopic measurements for extracting particle size distribution

Information on the particle size distribution (PSD) of a particulate system is contained in the turbidity measurements. To obtain the PSD from such measurements, the inversion of a Fredholm integral of the first kind is required. To solve this ill-posed problem Tikhonov regularization technique can be used. The central issue in every regularization problem is the choice of an effective regularization parameter and to perform such a choice in a complete automatic way without any inputs from the operator. In this work the PSD of polystyrene spheres in water are estimated from turbidity measurements using a CARY5000 spectrometer. Two formulations for the search of the regularization parameter based on cross-validation are compared and a two-step process integrating the two formulations is investigated.