3D Reconstruction of Placenta: SFM-MVS Technology for Large-scale Storage of Postpartum Placenta

Placental macroscopic assessment is vital for evaluating pregnancy outcomes, but is hindered by physical storage challenges and complex traditional 3D techniques. We present an efficient method using Structure from Motion (SfM) and Multi-View Stereo (MVS) to reconstruct 3D placental models from routine videos. Our pipeline extracts key frames, computes camera poses and a sparse point cloud via SfM, refines geometry with MVS, and generates a final model through Poisson reconstruction. Experiments demonstrate a reconstruction accuracy of 0.24 cm RMSE, meeting clinical standards. The method outperforms NeRF and 3DGS in structural completeness and computational efficiency. This approach enables digital preservation and retrospective analysis of placental macro-features, offering a practical tool for advancing placental pathology research.