Texture-based foam segmentation and analysis

A variety of polymers or surfactant mixtures or aeration of a number of liquids could generate mixtures of liquid and foam. Therefore, characterizing the properties of liquid/foam mixtures has important applications in the chemical process industry. The lack of a robust automated method for characterization within limited time and with high accuracy, however, has made this task difficult. In this work, we propose a new method based on image analysis for quantifying the geometric and statistical properties of such liquid/foam mixtures using images captured by an optical camera. The method can reliably achieve automated segmentation of liquid and foam layers. It can also find the boundaries of individual bubbles in the foam layer. At first, the region of interest, the foam region, is segmented from the input raw image. Then, the foam region is partitioned into two types of subregions, namely, loose foam or dense foam, according to local texture feature analysis. In the next step, to segment bubbles within the foam to obtain quantitative characterization, we apply two image processing algorithms, namely, Canny edge detection and K-means clustering, each specific to a different type of foam (loose or dense). The results show that the proposed automated segmentation and characterization method is robust and applicable to the processing of foam/liquid mixtures under many conditions.