Machine Learning Based Automatic Concrete Microstructure Analysis: A Study on Effect of Image Magnification

摘要

The scanning electron microscopy (SEM) images are commonly used to understand the microstructure of the concrete. Many researchers have adopted the image processing techniques for the microstructure analysis, but little has been studied on how the magnification of the SEM images influence the accuracy of analysis. Therefore, this paper presents a machine learning (ML) based framework to study the effect of SEM image magnification on degree of hydration measurement. In this study, the authors looked into the impact of magnification of SEM images on the model training, accuracy, and degree of hydration measurement using two scenarios. First, the image segmentation was performed using a classifier of specific magnification, and then a common classifier is trained using the image of different magnification. The preliminary results show that there is no significant effect of magnification on model training and accuracy. However, it has a significant impact on the degree of hydration measurement.