Comparative verification of 3D reconstructed point cloud data: comparison of AI estimation and multi-view stereo methods using Nerfstudio and Metashape

被引：0

作者：

Tsukamoto, Yuta ^{[1
]}

Hayashi, Atsushi ^{[2
]}

Tokuda, Kenichi ^{[2
]}

Kochi, Nobuo ^{[2
,3
]}

机构：

[1] Inst Sci Tokyo, Dept Mech Engn, Tokyo, Japan

[2] Natl Agr & Food Res Org, Res Ctr Agr Robot, 1-31-1 Kannondai, Tsukuba, Ibaraki 3050856, Japan

[3] Chuo Univ, R&D initiat, Tokyo, Japan

来源：

SICE JOURNAL OF CONTROL MEASUREMENT AND SYSTEM INTEGRATION | 2025年 / 18卷 / 01期

关键词：

3D reconstruction; NeRF; multi-view stereo; photogrammetry; point cloud; PHOTOGRAMMETRY;

D O I：

10.1080/18824889.2025.2497600

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

We investigated the feasibility of using a 3D point cloud estimated by a neural radiance field (NeRF). We evaluated the quality of point clouds reconstructed with NeRF and multi-view stereo. The point cloud reconstructed by NeRF contains substantial noise and shows a wavy surface even on a flat surface. Therefore, a point cloud generated by NeRF is not appropriate as a highly accurate 3D model with the size and shape of an actual object. However, it is effective for applications that do not require highly accurate point clouds, such as the detection and 3D modelling of objects with few feature-point surfaces or optically transparent objects, which have been difficult to achieve using conventional methods.

引用

页数：6

共 18 条

[1] Reconstructing Textureless Objects Image Enhancement for 3D Reconstruction of Weakly-Textured Surfaces [J].

Aldeeb, Nader H. ;

Hellwich, Olaf .

PROCEEDINGS OF THE 13TH INTERNATIONAL JOINT CONFERENCE ON COMPUTER VISION, IMAGING AND COMPUTER GRAPHICS THEORY AND APPLICATIONS (VISIGRAPP 2018), VOL 5: VISAPP, 2018, :572-580

[2]

[Anonymous], Agisoft Metashape

[3] Evaluation of accuracy of as-built 3D modeling from photos taken by handheld digital cameras [J].