Image Segmentation and Analysis for Microstructure and Property Evaluations on Ti–6Al–4V Fabricated by Selective Laser Melting

Abstract

The selective laser melting could be employed in fabrication of near-net shape products for airplane and biomedical applications from Ti–6Al–4V alloy, which is difficult-to-process material. In this method, the localized laser irradiation forms the unique Ti–6Al–4V microstructures which correspond to the laser scanning patterns and local thermal history as it could be observed from sample cross-sections with OM or SEM. In this study, the effects of heat treatments on mechanical properties of Ti–6Al–4V samples produced by selective laser melting are discussed based on quantitative analysis of microstructures with image processing and machine learning tools. It was found that microstructures of heat-treated samples retained their original morphologies and secondary α phase precipitated regularly at β grain boundaries with increased treatment time. These microstructures were appropriately segmented and classified. Each α particle geometrical characteristics were successfully extracted and evaluated by image analysis. Importantly, the hardness of the heat-treated samples was lower compared to that of as-built ones and it tended to increase with the area fraction of α phase, the α particle width, and the nearest neighbor distance between α particles.

Fig. 2 Scheme of image analysis (a) SEM images, (b) phase classification using machine learning with RF algorithm, (c) classified images, (d) α particles segmentation at k = 0.7, (e) ellipse approximation and (f) NND between α particles.