Port local pit rapid detection method based on unmanned aerial vehicle laser point cloud data
A rapid detection method for localized craters in ports using UAV laser point cloud data, combined with semantic segmentation and geometric feature analysis, solves the problems of detection hazards during port damage and low efficiency of multi-source data fusion. This method achieves high-precision and efficient port damage detection, especially for the identification of small-scale craters and boundary areas.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANJING UNIV
- Filing Date
- 2026-05-28
- Publication Date
- 2026-06-26
AI Technical Summary
Close-range measurement during port damage presents risks and low efficiency in multi-source data fusion processing, making it difficult to achieve high-precision and high-efficiency damage detection. In particular, in complex port environments, the detection of small-scale pits or boundary areas is prone to missed or false detections.
A rapid detection method for localized pits in ports based on UAV laser point cloud data is adopted. The method uses a 3D point cloud semantic segmentation network for preprocessing, combined with geometric feature analysis and sliding window block processing, to achieve accurate extraction of semantic point clouds on the dock surface and efficient processing of large-scale point cloud data.
It improves the accuracy and efficiency of port damage and crater detection, can simultaneously meet the detection needs of damage of different scales, reduces computational complexity, and provides accurate basic data support for port emergency repair and reopening plans.
Smart Images

Figure CN122289273A_ABST