Detection of Ship Wakes in Dynamic Sea Surface Video Sequences: A Data-Driven Approach

In order to enhance the detection of maritime vessel targets, considering the causal relationship between the motion of vessels and their wakes, as well as the characteristics of ship wakes such as large diffusion range and distinctive features, this paper proposes a data-driven method based on Dynamic Mode Decomposition (DMD) for detecting and analyzing ship wakes in sea surface videos. The method, named Multi-dimensional Dynamic Mode Decomposition (MDDMD), segments the video sequence into smaller blocks and analyzes them at various resolution levels, effectively addressing the data analysis issues of large and complex systems. The MDDMD algorithm not only extracts key dynamic features but also reveals the intrinsic structure of the system at different scales, providing new perspectives for the in-depth understanding of nonlinear systems. Comparative experimental results with existing DMD and PCA algorithms demonstrate that the MDDMD algorithm has higher accuracy and robustness in ship wake detection. This study offers valuable insights for ship wake detection under complex maritime conditions and holds potential for practical application in the field of maritime surveillance.