This paper describes how the formation of particle systems can be investiga
ted by using computer vision techniques, namely effective exact dilations a
nd multiscale skeletonization. By assuming that all particles started to gr
ow at the same time and had the same growth rate, a hypothesis that can be
validated later, it becomes possible to infer the shape and distribution of
the particles in the initial system configuration. Multiscale skeletons ar
e obtained for each individual particle, and a suitable overall spatial sca
le, i.e. those best approximating the initial configuration, is determined
as that immediately before the smallest particle fades. The thus obtained s
keletons provide an estimate of the shape and position of the initial parti
cles before uniform expansion. By using exact dilations, a generalized Diri
chlet tessellation reconstruction of the particle system is obtained from t
hese skeletons, and the similarity between this reconstruction and the orig
inal system can be used to validate the assumptions about the growth condit
ions. The proposed methodology is illustrated with respect to KCl polycryst
alline thin films.