Analysis of microscopic images using image processing to measure homogeneity of metal powder mixture samples

Abstract

This paper presents automated image analysis frameworks that leverage Python's computer vision libraries, including OpenCV and Matplotlib, to analyze and characterize color-shade distributions in microscopic images of metal powder mixtures. The proposed framework segments the images into multiple regions of interest (ROIs), calculates and visualizes color histograms, and extracts quantitative information about the color distribution within each ROI. The developed image processing framework measures the homogeneity of the powder mixture by surface analysis of the image of samples. This article demonstrates the efficacy of these frameworks on a set of microscopy images, highlighting their potential to measure the homogeneity of the metal powder mixture. Homogeneity is a crucial factor in material science, as it directly impacts the performance and properties of the resulting material. The developed technique is economical compared to available methods to predict the homogeneity of metal powder mixtures, such as Tomographic and Spectroscopic analyses. It is an effective technique to measure the homogeneity of metal powder mixtures exhibiting different colors or shades of color in computer vision. In this work, the homogeneity is analyzed for the powder mixtures consisting of 80% aluminum (Al) and 20% copper (Cu), 90% aluminum (Al) and 10% copper (Cu) and 90% silicon carbide (SiC) and 10% aluminum (Al) using processing of microscopic color images.