Three-dimensional reconstruction of green water on deck based on SLoFTR matching

Accurate measurement of green water on deck is important for assessing wave-induced loads, but conventional gauges provide only pointwise information. This study proposes a stereo-vision method based on SIFT-anchor-guided LoFTR (SLoFTR) for spatially resolved reconstruction of green water on deck. High-confidence SIFT correspondences are used as anchors to estimate epipolar and spatial-support priors, which are introduced into LoFTR matching to improve correspondence estimation in weak-texture regions affected by reflection and foam. The resulting matches are triangulated and transformed into a deck-fixed coordinate system through stereo and zero-plane calibration. Free-running model tests were conducted, and reconstructed elevations were validated at three ultrasonic-gauge locations. Compared with vanilla LoFTR, SLoFTR increased the average numbers of matches and inliers per stereo image pair by 40.9% and 30.5%, respectively, while maintaining a comparable median Sampson distance. Across the three validation points, the Pearson correlation coefficient increased from 0.715 to 0.824 to 0.797-0.946, and the mean squared error decreased by 14.0%-72.3%. SLoFTR also produced more continuous sectional profiles and full-field surfaces. These results demonstrate its potential as a practical non-contact method for spatially resolved measurement of green water on deck.

Authors

Institutions

Publication Details

Journal
Ocean Engineering
Published
2026-10-05
DOI
https://doi.org/10.1016/j.oceaneng.2026.128472
Primary Topic
Advanced Vision and Imaging
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Three-dimensional reconstruction of green water on deck based on SLoFTR matching

GU Xue-kang, Bo Lan, Han-Yu Sun, Ying-Dong Chen et al.
Ocean Engineering
Advanced Vision and Imaging
article

Three-dimensional reconstruction of green water on deck based on SLoFTR matching

GU Xue-kang, Bo Lan, Han-Yu Sun, Ying-Dong Chen, Nan Zhao, Zhen-Yuan Li
article en

Abstract

Accurate measurement of green water on deck is important for assessing wave-induced loads, but conventional gauges provide only pointwise information. This study proposes a stereo-vision method based on SIFT-anchor-guided LoFTR (SLoFTR) for spatially resolved reconstruction of green water on deck. High-confidence SIFT correspondences are used as anchors to estimate epipolar and spatial-support priors, which are introduced into LoFTR matching to improve correspondence estimation in weak-texture regions affected by reflection and foam. The resulting matches are triangulated and transformed into a deck-fixed coordinate system through stereo and zero-plane calibration. Free-running model tests were conducted, and reconstructed elevations were validated at three ultrasonic-gauge locations. Compared with vanilla LoFTR, SLoFTR increased the average numbers of matches and inliers per stereo image pair by 40.9% and 30.5%, respectively, while maintaining a comparable median Sampson distance. Across the three validation points, the Pearson correlation coefficient increased from 0.715 to 0.824 to 0.797-0.946, and the mean squared error decreased by 14.0%-72.3%. SLoFTR also produced more continuous sectional profiles and full-field surfaces. These results demonstrate its potential as a practical non-contact method for spatially resolved measurement of green water on deck.

Ocean EngineeringVol. 368
China Ship Scientific Research Center (CN)
Openalex Percentile: Top 14%
Advanced Vision and Imaging
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.