Automatic 3D reconstruction of manifold meshes via delaunay triangulation and mesh sweeping

Abstract

In this paper we propose a new approach to incrementally initialize a manifold surface for automatic 3D reconstruction from images. More precisely we focus on the automatic initialization of a 3D mesh as close as possible to the final solution; indeed many approaches require a good initial solution for further refinement via multi-view stereo techniques. Our novel algorithm automatically estimates an initial manifold mesh for surface evolving multi-view stereo algorithms, where the manifold property needs to be enforced. It bootstraps from 3D points extracted via Structure from Motion, then iterates between a state-of-the-art manifold reconstruction step and a novel mesh sweeping algorithm that looks for new 3D points in the neighborhood of the reconstructed manifold to be added in the manifold reconstruction. The experimental results show quantitatively that the mesh sweeping improves the resolution and the accuracy of the manifold reconstruction, allowing a better convergence of state-of-the-art surface evolution multi-view stereo algorithms.

Robust moving objects detection in lidar data exploiting visual cues

“…Experiments show that ground plane removal and scan comparison discretization improve on precision with respect to current state-of-theart with a speed-up in the execution thanks to the use of an efficient indexing data structure. As a future work we aim at applying the proposed approach to an existing urban reconstruction method [19] and refine it with [20] in order to obtain a 3D urban map without moving objects, while still improving on the speed up of the visual pipeline of our proposal.…”

Applying photogrammetry to reconstruct 3D luminance point clouds of indoor environments

“…From the 1990s, digital photogrammetry enabled automatic image measurements, camera calibration and exterior orientation (Haggrén & Niini, 1990;Heikkilä & Silven, 1997;Lowe, 1999;Pollefeys, Koch, & Van Gool, 1999;Stathopoulou, Welponer, & Remondino, 2019). Current photogrammetric software can automatically reconstruct 3D mesh models (Furukawa, Curless, Seitz, & Szeliski, 2009;Jancosek & Pajdla, 2011;Romanoni, Delaunoy, Pollefeys, & Matteucci, 2016). For indoor modelling, various photogrammetric methods have been used, such as 3D mapping systems (El-Hakim, Boulanger, Blais, & Beraldin, 1997), videogrammetry-based 3D modelling (Haggrén & Mattila, 1997), structured indoor modelling (Ikehata, Yang, & Furukawa, 2015), cloud-based indoor 3D modelling (Ingman, Virtanen, Vaaja, & Hyyppä, 2020) and other indoor measuring methods (Georgantas, Brédif, & Pierrot-Desseilligny, 2012;Lehtola et al, 2017;León-Vega & Rodríguez-Laitón, 2019).…”

Surface reconstruction from a sparse point cloud by enforcing visibility consistency and topology constraints

“…This is useful for downstream processing such as smoothing, refinement and texture mapping ( [42,5]), which require low topological noise. An example is the dense stereo refinement: the surface evolves in 3D such that it minimizes a photo-consistency function without changing the genus ( [12,40,33]). The lower the topological noise, the better the expected convergence.…”