Image-Based Analysis and Visualization of Hull Deformation Regions: An Integrated Approach Using Depth Estimation and Curvature Analysis

Detecting deformations in ship hull structures is crucial for maritime safety and ship maintenance. Conventional deformation detection techniques tend to be costly, time-consuming, and often require complex equipment. This study proposes an efficient and accurate method for detecting ship hull deformations using a monocular camera. The method involves capturing images of the deformed hull surface, estimating depth through image analysis, and converting the estimated depth data into a point cloud.BR/The point cloud is then used to establish a reference plane from the projected points, and deviations and curvature relative to this plane are measured to identify deformed areas. The proposed technique demonstrates its validity by comparing the results with actual deformation regions. Despite relying on only a single image, the method shows high accuracy and the potential for detecting deformations across various ship hull types. The efficiency of deformation detection is further improved by image analysis techniques that calculate the curvature of the deformed hull. Furthermore, the method shows the potential to handle unlabeled data, providing a significant advantage over deep learning models that typically require large datasets, as it can function effectively with just one image.