Quantitative Imaging of Colloidal Structures

Abstract

Quantitative analysis of microscopy images is an essential tool in the study of colloidal materials, but extracting precise structural information can be hindered by unfavorable or inhomogeneous image statistics and complex object shapes, such as polydispersity, anisotropy, and asymmetry. Here, we address these challenges with image processing and analysis methods that ensure accurate binarization of complex images, followed by algorithms for extracting structural features of colloidal aggregates and suspensions. Metrics grounded in fundamental morphological features of binary objects are defined to describe the dimensions, surface structure, alignment, orientation, and distribution of objects in an image. The approach is particularly suitable for data sets where manual labeling is impractical, but deep learning methods are not feasible. The methodology is validated on a diverse set of video micrographs of self-assembled colloidal clusters. The proposed methods characterize suspension structures across multiple length scales, demonstrating high accuracy and reproducibility. Accessible Python scripts are provided to facilitate data analysis, making the workflow broadly applicable to microscopy data evaluation in numerous areas of colloid science.