Multi View Stereo
Multi-view stereo (MVS) aims to reconstruct a 3D model of a scene from multiple 2D images by identifying corresponding points across different viewpoints. Current research emphasizes improving accuracy and efficiency, particularly in challenging scenarios like textureless surfaces or sparse data, through advancements in deep learning architectures such as transformers and neural radiance fields (NeRFs), as well as refined cost aggregation and geometric consistency methods. These improvements are driving progress in applications ranging from autonomous driving and robotics to cultural heritage preservation and medical imaging, where accurate 3D scene reconstruction is crucial.
Papers
TransMVSNet: Global Context-aware Multi-view Stereo Network with Transformers
Yikang Ding, Wentao Yuan, Qingtian Zhu, Haotian Zhang, Xiangyue Liu, Yuanjiang Wang, Xiao Liu
IB-MVS: An Iterative Algorithm for Deep Multi-View Stereo based on Binary Decisions
Christian Sormann, Mattia Rossi, Andreas Kuhn, Friedrich Fraundorfer