ROV for Underwater Observation in Low-Transparency Waters

• Construct an underwater spatially adaptive discrete light field guided by acoustics, and integrate a multimodal controllable light-source array to achieve precise energy delivery.
• Develop a technology for coordinated control of light source energy distribution and imaging area to mitigate the effects of scattering and attenuation of light propagation in turbid water.
• By integrating deep learning with environmental parameters, we have developed an underwater image restoration and enhancement algorithm that adaptively estimates the point spread function and medium transmittance under varying water‑quality conditions. This enables correction of forward scattering and removal of backscattering, ultimately improving image clarity and target‑recognition performance in low‑transparency waters.

Technical Performance Specifications

Indicator Name	Indicator parameters
Optical detection range	≥1.5 times visibility
Acoustic detection range	≥50 meters
Acoustic field of view	Horizontal ≥ 130°, Vertical ≥ 20°
Acoustic detection blind spot	≥0.2 meters

Underwater image processing algorithm based on deep learning and environmental parameters;

By employing machine learning–based optimization algorithms, the adverse impact of effective information loss caused by image degradation during underwater dynamic operations on operational performance and efficiency has been effectively mitigated.

A novel underwater detection image-processing method based on multiphysics-field data has been developed.

 

Underwater Visualization System: Marine Engineering Application Scenarios