Research on Floating Ice Recognition and Wave Reconstruction Using Binocular Vision

With the increasing demand of Arctic shipping routes, the navigation safety of polar ships has attracted much attention. Floating sea ice and waves pose potential threats to ships, so how to effectively identify and respond to these natural phenomena is particularly important. This paper proposes a computer vision-based method to identify moving sea ice in waves and perform a three-dimensional reconstruction of the wave front. First, the binocular camera is used to collect video data as input, and the antipolar line where the possible matching points are located is determined through the antipolar constraint, so as to reduce the search space and improve the matching efficiency. Then, the LK sparse optical flow method and sgbm method are used to identify the moving floating ice model, and the underwater point cloud data and water surface height data are obtained. The LK sparse optical flow method captures the moving track of the floating ice, while the sgbm method provides high-precision stereoscopic matching results. Finally, the data was visually analyzed, and the results showed that the method was able to quickly identify the sea ice and accurately measure and reconstruct the height of the three-dimensional waves.