Surface Reconstruction from Point Clouds: A Survey and a Benchmark

Abstract

Reconstruction of a continuous surface of two-dimensional manifold from its raw, discrete point cloud observation is a long-standing problem in computer vision and graphics research. The problem is technically ill-posed, and becomes more difficult considering that various sensing imperfections would appear in the point clouds obtained by practical depth scanning. In literature, a rich set of methods has been proposed, and reviews of existing methods are also provided. However, existing reviews are short of thorough investigations on a common benchmark. The present paper aims to review and benchmark existing methods in the new era of deep learning surface reconstruction. To this end, we contribute a large-scale benchmarking dataset consisting of both synthetic and real-scanned data; the benchmark includes object-and scene-level surfaces and takes into account various sensing imperfections that are commonly encountered in practical depth scanning. We conduct thorough empirical studies by comparing existing methods on the constructed benchmark, and pay special attention on robustness of existing methods against various scanning imperfections; we also study how different methods generalize in terms of reconstructing complex surface shapes. Our studies help identify the best conditions under which different methods work, and suggest some empirical findings. For example, while deep learning methods are increasingly popular in the research community, our systematic studies suggest that, surprisingly, a few classical methods perform even better in terms of both robustness and generalization; our studies also suggest that the practical challenges of misalignment of point sets from multi-view scanning, missing of surface points, and point outliers remain unsolved by all the existing surface reconstruction methods. We expect that the benchmark and our studies would be valuable both for practitioners and as a guidance for new innovations in future research. We make the benchmark publicly accessible at https://Gorilla-Lab-SCUT.github.io/SurfaceReconstructionBenchmark.

Single-sized spheres on surfaces (S4)

“…For more densely sampled areas, a smaller splat size guarantees the creation of vertices closer to the sampling points. In our preliminary experiments, we saw that in high-curvature regions, smaller splats have small deviation from the surface, while larger splats [13]. The result 7a is obtained with global splat sizes.…”

“…In our experiments, we focus on the reconstruction of real-world scan data. For this, we turn to 20 scanned objects provided as part of a surface reconstruction benchmark [13]. Here, we concentrate on high-resolution scans obtained by an OKIO 5M scanning device, resulting in 330k to 2,000k points per surface after 20 shots.…”

“…The shots are registered and do come with a normal field. All data sets are publicly available, see the repository of [13]. Out of the 20 point clouds, we used 19 as they are provided in the repository.…”

“…This filter repeatedly applies edge flip, collapse, relax, and refine operations. We run 3 iterations with a target edge length of 0.2 in absolute world units for the input of [13]. In Tables 1 to 4, we report both the results of the comparison algorithms and the result after these have been remeshed, indicated by "(Re)".…”

“…Other user requirements are, for instance, a result mesh of high quality, i.e., consisting of triangles of length close to uniformity and vertices of degree close to 6, or low computational costs. For an overview of surface reconstruction algorithms, we refer to a recent surface reconstruction survey [13].…”