基于三维激光雷达的沿海泊位码头识别算法  

Wharf Identification Algorithm Based on 3D LiDAR

作　　者：朱飞祥[1] 王少博[1] 邢胜伟[1] 那文杰 ZHU Feixiang;WANG Shaobo;XING Shengwei;NA Wenjie(Navigation College,Dalian Maritime University,Dalian 116026,China)

机构地区：[1]大连海事大学航海学院,辽宁大连116026

出　　处：《中国航海》2020年第3期75-78,96,共5页Navigation of China

基　　金：国家重点研发计划(2018YFB1601505);辽宁省航运联合基金(2020-HYLH-28);中央高校基本科研业务费专项资金(3132019313);国家自然科学基金(51709032)。

摘　　要：针对海上水面自主船舶(Maritime Autonomous Surface Ship,MASS)在港内航行或自动靠离泊过程中面临的近距离典型障碍物自动检测和识别的问题,提出一种运用三维激光雷达(Light Detection And Ranging,LiDAR)获取物标几何特征辨识沿海码头的算法。基于开源的点云库(Point Cloud Library,PCL),采用统计滤波器技术对原始点云数据去除噪声,提出基于随机采样一致算法(Random Sample Consensus,RANSAC)的码头聚类分割模型,构建基于主成分分析的有向包围盒(Oriented Bounding Box,OBB),计算获取码头的长度、宽度和表面水平性的几何特征,实现对沿海典型码头的识别。实船试验结果表明:该算法正确可靠,具备一定的鲁棒性,可为利用LiDAR实现船舶港内自主避障和自动靠离泊研究提供参考。The ability of detecting and identifying obstacles in short range is critical for MASS(Maritime Autonomous Surface Ship) to navigate and berth in an unknown environment. The wharf identification algorithm based on 3 D LiDAR(Light Detection and Ranging) is a design fitting with that needs. Based on the open source PCL(Point Cloud Library), the outliers are removed from the original point cloud data by means of the statistical filtering technology. The wharf clustering segmentation model based on the RANSAC(Random Sample Consensus) is proposed. The OBB(Oriented Bounding Box) based on Principal Components Analysis is constructed for calculating the length, the width and the surface level of the wharf, so as to identify the typical wharfs along the coast. Experiments using 3 D LiDAR on board a merchant ship while trying to berth automatically indicate that the algorithm is reliable and robust.

关 键 词：激光雷达 码头 点云数据 聚类分割 

分 类 号：TN958.98[电子电信—信号与信息处理] U676.1[电子电信—信息与通信工程]