Combining machine learning and Mie theory to simplify particle characterization inside microchannels

Abstract

In recent decades, particle characterization has been one of the most widely used achievements. The article presents a method to simplify the setup using unsupervised machine learning techniques, such as K-means, K-medoids, and Hierarchical clustering. Utilizing these three methods together, our approach can accurately measure particle diameter with a precision of 0.1 μm and a refractive index of 0.001 using only a laser and a camera without the need for complex alignment of components. Furthermore, our method is capable of separating scattered signal images from background images.