"Aircraft Fuselage Dataset Acquired by a Mobile Structured Light Scanner"
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
"Accurate multi-view 3-D measurement of large-scale, feature-scarce surfaces remains challenging because limited inter-view observability readily causes cu mulative drift, while dense global refinement over high resolution scans is often computationally prohibitive for practical real-time deployment. To address these chal lenges, this article presents a fiducial-assisted multi-view 3-D measurement framework driven by a novel dual-tier optimization strategy. Specifically, in the first tier, we for mulate a probabilistic bundle adjustment (BA) model that jointly optimizes camera and fiducial poses by integrating subpixel boundary, stereo-depth, and normal constraints. Furthermore, robust losses and covariance weighting are utilized to effectively suppress registration errors. In the second tier, a boundary-constrained dense global refine ment couples fiducial-geometric residuals to suppress drift and subtle stratification while preserving dense surface consistency. To ensure computational efficiency, the dense optimization is linearized via Gauss-Newton, and the result ing high-dimensional sparse normal equations are solved via a GPU-accelerated preconditioned conjugate gradient method. On-site comparative experiments using a mobile structured light scanner demonstrate the effectiveness of the proposed dual-tier optimization strategy. With an av erage pairwise registration time of 56 ms, the proposed method achieves measurement RMSEs of 0.828 mm and 0.925 mm on the aircraft fuselage and wing, respectively, which approach the scanner\u2019s nominal 0.600 mm calibra tion accuracy."
Fields of Study & Subjects
Optical measurement and interference techniques
3D Surveying and Cultural Heritage
Advanced Vision and Imaging
Publication details
- Journal / source
- IEEE DataPort
- Publication date
- 30 March 2026
- Language
- EN
- Document type
- Dataset