Analysis of the microcapsule structure based on machine learning algorithm

Abstract

For the microcapsules preparation process, the mechanism is estimated by structure reformation during the preparation process since diameter and wall thickness drastically changed. Microstructures are recently studied by machine learning techniques. The Hough transformation algorithm is used by other researchers for the preparation of the microcapsules but it is difficult to determine the mechanism by using only a diameter change of the microcapsules. Therefore, one additional way to establish the mechanism is the analysis of the formation of the microcapsule structure. In this study, The Hough transformation algorithm was used for the image segmentation, the simple feature extractions were checked and the support vector machine and the k-nearest neighbors algorithm were used as classifiers in order to analyze the structure of the microcapsules which were prepared by solvent evaporation method from a solid in oil in water, S/O/W, emulsion system. The structural distribution was analyzed by the developed detection method. The microcapsules had a specific structural distribution which are monocore, multicore, and other aggregated structures. The structural distribution was changed by the preparation condition. The monocore structure was dominant by increasing in the amount of water soluble solid particles added in the organic phase.