Real-time hatch detection during unloading of dry bulk carriers with side rolling hatchcover

Accurately positioning the hatch during the loading and unloading of bulk cargo carriers is a critical factor that impacts safety and efficiency in cargo handling processes. Current methods, which involve reconstruction followed by hatch opening detection, are time-consuming and often lead to inaccurate positioning due to variations in cargo weight and sea conditions. To overcome these limitations, we propose a hatch recognition algorithm based on point cloud generation and image contour extraction to meet the rapid detection requirements of large cargo ships. We employ histogram filtering to eliminate point cloud data originating from cargo inside the hatch during real-time scanning, thereby determining the plane where the hatch opening is situated. The point cloud is then vertically projected onto the hatch cover surface and transformed into a binary image, ensuring precise hatch positioning. Experimental validation demonstrates the reliability of this algorithm, with an error rate of less than 1.5% and a processing time of under 3 seconds. This method enables real-time guidance for loading and unloading ships, contributing to enhanced cargo handling speed and safety. Furthermore, this technology holds the potential for successful application in bulk cargo handling operations at freight terminals in the future.