Investigation of Sensors Related 3D-Mapping Algorithm for Monocular Video based Underwater Vehicle

In recent years, underwater operations have increasing requirements for emerging discovery jobs in the sea. Therefore, it is necessary to improve the safety and productivity using state-of-the-art ICT technology such as 3D-mapping on underwater vehicles. Although some previous 3D-mapping technologies which have achieved excellent performance in air, it is still difficult to achieve high-precision underwater 3D-mapping. In this study, we investigate an excellent sensor related 3D-mapping algorithm, namely NeuralRecon, which realized real-time generation of 3D-mapping in the air. All the related sensors used by NeuralRecon are analyzed to confirm its underwater performance. The investigation results shows that the sensors can be categorized into three types and should be used based on different conditions.