An experimental investigation of the time-dependent spatial distribution of droplet concentration in a ‘box’ of stationary homogeneous and isotropic turbulence without mean flow was performed for polydispersed droplet clouds with a wide range of mean droplet diameters and droplet size distributions, characterized by a representative Stokes number, based on the droplet arithmetic diameter and the Kolmogorov timescale of the flow, varying between 0.2 and 6, and for turbulent Reynolds numbers of 98, 127, 147 and 235. A novel morphological tracking scheme was used to provide temporal tracking of droplet clusters according to cluster morphological features. The tracking scheme was based on Voronoï space tessellation, multivariate kernel density estimation and mean shift feature space tracking. The developed tracking scheme was used to establish the relationship between the morphological features of droplet clusters, i.e. number of clusters per unit area and cluster lengthscales, and the velocity of droplet clusters. The timescale of the droplet clusters increased with both the turbulent Reynolds number and the Stokes number, but its dependence was stronger on the turbulence level. In addition, the effect of filling factors of turbulent flows by turbulent structures on droplet clustering was discussed, which led to suggestions of potential flow control methods to mitigate droplet preferential concentration.