<p>Reliable density measurements of molten metals are crucial in understanding the evolution and formation of their phase structures. However, they are very challenging with container methods due to the possible parasitic reactions, wetting and meniscus formation, biased volume estimates, <i>etc</i>. Electrostatic levitation (ESL) is a containerless technique that alleviates these limitations while keeping the molten metals close to spheres for accurate volume measurements. This paper uses image analysis techniques to perform single-view bidirectional diameter extraction of non-ideal spherical specimens in ESL through ellipse fitting and Legendre polynomial fitting. More importantly, the proposed single-camera single-view workflow can reach an accuracy level comparable to dual-camera 3D reconstruction approaches reported in the literature, while requiring only one optical access. After image inversion, edge detection, and contour extraction, the droplet contour is fitted using two models: an ellipse and a Legendre polynomial curve. From these fits, the horizontal and vertical diameters are obtained and used to estimate the droplet volume. To make the workflow practical for large datasets, we implemented the entire pipeline in code and developed a fully automated batch program. It processes all images in a folder without manual intervention, generates annotated output images for each frame, and exports the key measurement results to an Excel file. Tests on heating–cooling images of levitated zirconium show that both fitting methods work reliably, while ellipse fitting is typically more stable when the contour is clean and continuous. Overall, the proposed approach improves the efficiency and consistency of diameter/volume extraction and supports high-throughput analysis for thermophysical property characterization.</p>