RepC-MVSNet: A Reparameterized Self-Supervised 3D Reconstruction Algorithm for Wheat 3D Reconstruction

The application of 3D digital models to high-throughput plant phenotypic analysis is a research hotspot nowadays. Traditional methods, such as manual measurement and laser scanning, have high costs, multi-view, unsupervised reconstruction methods are still blank in the field crop research. It challenging obtain high-quality surface feature composition for reconstruction. In this paper, we propose wheat point cloud generation method based on SfM MVS using sequential images. Firstly, camera intrinsics extrinsics were estimated structure-from-motion system with maps, which effectively solved problem location design. Secondly, proposed ReC-MVSNet, integrates heavy parametric structure into network, overcoming difficulty capturing complex features via traditional model. Through experiments, it was shown that achieves non-invasive realistic objects, accuracy model improved by nearly 43.3%, overall value 14.3%, provided new idea development virtual digitization.