Developing a program for quantification and characterization of microstructure: Application to BSE images of geopolymer recycled concrete

Abstract

This study aims to achieve efficient, accurate, and automated identification of phase features in the backscattered electron (BSE) images of geopolymer recycled concrete (GRC), enhancing the understanding of its microstructural characteristics and facilitating the investigation of the relationship between microstructure and macroscopic mechanical performance. A system for phase recognition and quantitative analysis of GRC, named CMAS, is developed. This system performs quantitative analysis and calculations based on various microstructural images. The program first automatically removes noise, enhances contrast, and corrects image defects. It then applies the multi-Otsu thresholding method and binarization to segment and identify the phase components. Finally, the system employs connectivity analysis and geometric operations to derive ten microstructural parameters, including area, length, orientation, perimeter, fractal dimension, uniformity coefficient, and curvature coefficient. Additionally, the system generates corresponding rose diagrams. The proposed system can automatically process large numbers of images, significantly improving data processing efficiency and user-friendliness. All operations are non-destructive to the images, making it a powerful tool for the analysis of material microstructural images and providing valuable reference for the development of similar image processing systems.