Abstract:We develop a phase-resolved wave field reconstruction method by the learning-based downsampling network for processing large amounts of inhomogeneous data from non-contact wave optical observations. The Waves Acquisition Stereo System (WASS) extracts dense point clouds from ocean wave snapshots. We couple learning-based downsampling networks with the phase-resolved wave reconstruction algorithm, and the training task is to improve the wave reconstruction completeness ratio CR. The algorithm first achieves initial convergence and task-optimized performance on numerical ocean waves built by the linear wave theory model. Results show that the trained sampling network can lead to a more uniform spatial distribution of sampling points and improve CR at the observed edge regions far from the optical camera. Finally, we apply our algorithm to a natural ocean wave dataset. The average completeness ratio is improved over 30% at low sampling ratios (SR∈[2−9,2−7]) compared to the traditional FPS method and Random sampling method. Moreover, the relative residual between the final reconstructed wave and the natural wave is less than 15%, which provides an efficient tool for wave reconstruction in ocean engineering.

What problem does this paper attempt to address?

The paper attempts to address the problem of how to efficiently and accurately reconstruct phase-resolved wave fields using remote optical observation data in ocean engineering. Specifically, the paper focuses on handling large amounts of non-uniformly distributed data through a learning-driven downsampling network to improve the completeness and accuracy of wave field reconstruction. Traditional methods face issues such as low efficiency and uneven distribution of sampling points when dealing with large-scale point cloud data, which can lead to wave information loss in edge areas and thus affect reconstruction accuracy. The proposed method aims to optimize the sampling strategy to reduce data volume while preserving wave field characteristics, thereby enhancing the performance of wave field reconstruction tasks. The main contributions of the paper include: 1. **Proposing a task-driven learning downsampling network**: This network can dynamically adjust the sampling strategy according to the needs of downstream tasks, ensuring that the sampled point cloud is more evenly distributed in space, thereby improving the completeness ratio (CR) of wave field reconstruction. 2. **Incorporating a linear wave theory model**: Numerical waves are generated using the linear wave theory model for training and evaluating the downsampling network, ensuring the network's effectiveness in practical applications. 3. **Experimental validation**: The effectiveness of the proposed method is validated through experiments on simulated datasets and actual wave datasets. Experimental results show that compared to traditional random sampling and farthest point sampling methods, the proposed method significantly improves the completeness ratio of wave field reconstruction at low sampling rates, and the relative residual between the reconstructed wave and the natural wave is less than 15%. In summary, by introducing a learning-driven downsampling network, the paper addresses the issues faced by traditional methods in handling large-scale non-uniformly distributed optical observation data, providing a new effective tool for wave field reconstruction in ocean engineering.

Task-Driven Learning Downsampling Network Based Phase-Resolved Wave Fields Reconstruction with Remote Optical Observations

Wavefront Reconstruction Based on Deep Transfer Learning for Microscopy.

Atmospheric Turbulence Phase Reconstruction via Deep Learning Wavefront Sensing

A Physics-Driven CNN Model for Real-Time Sea Waves 3D Reconstruction

Real-time Underwater 3D Reconstruction Based on Monocular Image

Machine learning for phase-resolved reconstruction of nonlinear ocean wave surface elevations from sparse remote sensing data

Improved Machine Learning Approach for Wavefront Sensing

Scale-Translation Equivariant Network for Oceanic Internal Solitary Wave Localization

A Noncontact Technique for Wave Measurement Based on Thermal Stereography and Deep Learning

Resolution-Free Point Cloud Sampling Network with Data Distillation.

Phase-resolved real-time forecasting of three-dimensional ocean waves via machine learning and wave tank experiments

Transfer learning in large-scale ocean bottom seismic wavefield reconstruction

Training Deep Networks with Only Synthetic Data: Deep-learning-based Near-Offset Reconstruction for (Closed-Loop) Surface-Related Multiple Estimation on Shallow-Water Field Data

Research on High-Resolution Reconstruction of Marine Environmental Parameters Using Deep Learning Model

Super-Resolution Image Reconstruction of Wavefront Coding Imaging System Based on Deep Learning Network

A Multilevel Sampling Method for Detecting Sources in a Stratified Ocean Waveguide

Learning a Compact Spatial-Angular Representation for Light Field

A transfer-learning-based windspeed estimation on the ocean surface: implication for the requirements on the spatial-spectral resolution of remote sensors

Wave (from) Polarized Light Learning (WPLL) method: high resolution spatio-temporal measurements of water surface waves in laboratory setups

A deep-learning reconstruction method for remote sensing images with large thick cloud cover

Dual-Attention-Based Wavelet Integrated CNN Constrained Via Stochastic Structural Similarity for Seismic Data Reconstruction