基于LiDAR点云的船载砂石体积自动测算系统

针对现有船舶运输砂石体积计算主要依赖人工方式称重和纸质运单流转、精度不高且效率很低的难题，采用基于八叉树的点云精简和聚类分析进行复杂环境下船舶点云的自动识别和提取；采用基于改进SK-4PCS （semantic-keypoint-based 4-points congruent sets）的粗配准和基于Point-to-Plane ICP （iterative closest point）的精配准进行空载和满载两期点云由粗到精的高效配准，设计实现了一种利用LiDAR （light detection and ranging）点云数据进行船载砂石自动测算的系统. 采集了江西南昌赣江运砂船数据进行实验验证，从扫描到出结果一艘船的砂石体积自动测算时间少于2 min，同船多次测算结果最大相对误差小于1.00%，砂石体积与实际真值相比的最大误差小于2.00%，满足实际业务需求. 

Automatic Volume Calculation System for Sand and Gravel Carried by Ship Based on LiDAR Point Cloud

The current volume calculation method of sand and gravel carried by ship mainly relies on manual weighing and waybill transfer, which has the problem of low precision and low efficiency. To deal with this, an automatic volume calculation system of sand and gravel carried by ship is developed on the basis of LiDAR (light detection and ranging) point cloud. It uses the cluster analysis and the octree-based point cloud reduction to realize automatic identification and extraction of the point cloud for the target ship in a complex environment. It adopts the coarse registration based on the improved SK-4PCS (semantic-keypoint-based 4-points congruent sets) and fine registration based on the point-to-plane ICP (iterative closest point) to realize the high-efficient registration of no-load and full-load point clouds from coarse to fine level. The experiment is conducted with the data from the gravel transport ships on the Ganjiang River in Nanchang, Jiangxi, showing that the automatic measurement and calculation time of sand and gravel volume for one ship is less than 2 min, the maximum relative error of multiple calculation results is less than 1.00%, and the maximum error between the volume calculation results and real values is less than 2.00%, which meet practical requirements.