Algorithm for hatch recognition of port ship loaders based on 3D laser point cloud

With the development of smart ports, research on unload and shipment system automatization is becoming increasingly important. Hatch recognition, as a crucial link in automated shipment systems, is a key factor influencing the efficiency and accuracy of subsequent loading operations. To address the issues of real-time performance and safety in bulk cargo automatic loading operations, an automatic hatch identification algorithm for port ship loaders is proposed based on 3D laser point clouds. Firstly, a kd-tree index data structure is constructed to ensure the efficiency of point cloud data retrieval, and omp multi-threading programming is adopted to increase the processing speed of point cloud data. Subsequently, the plane hypothesis clustering is used to select the maximum density plane subset for boundary extraction and plane fitting, calculating the deck normal vector, and outputting the comprehensive ship information. Finally, hatch recognition is performed through 2D image recognition and 3D point cloud recognition. Experimental results show that this algorithm can ensure the accuracy of hatch identification while reducing the time for hatch recognition, effectively improving the recognition efficiency.