摘要
三维扫描数据和工艺模型间的配准定位是船舶建造精度管理与控制的关键,采用常规的配准算法难以有效抑制粗差对位姿定位精度的影响。为实现快速、准确配准,给出合理的建造精度评价,提出一种基于最大相关熵的鲁棒性点云配准算法,以削弱粗差对配准精度影响。该算法的核心是将高斯核函数引入到配准学习中。通过构建以高斯核函数为核心的稳健估计模型,利用迭代策略反复测试关联参数。理论和实际算例验证了该算法的有效性和实用性,为船舶建造精度提高和后续装配提供参数。
Pose registration between the 3D scanning data and the theoretical model is the key to the management and control of ship construction accuracy.Conventional registration algorithms cannot effectively suppress the influence of gross errors on the positioning accuracy.In order to realize fast and accurate registration,this paper proposes a robust registration algorithm for point clouds based on maximum correntropy criterion to weaken the influence of gross errors on the registration accuracy.Core of the proposed algorithm is Gaussian kernel function and is introduced into registration learning in this study.We establish a robust estimation function with Gaussian kernel function,and optimize the objective function by use of an iterative strategy for test and adjust the incidence matrix.Both theoretical and hull block experimental results confirmed the validity and practicability of the algorithm,providing a quantitative data basis for the verification of models and subsequent assembly.
作者
杨贵强
李瑞
刘玉君
汪骥
周玉松
YANG Guiqiang;LI Rui;LIU Yujun;WANG Ji;ZHOU Yusong(School of Naval Architecture,Dalian University of Technology,Dalian 116024,China;State Key Laboratory of Structural Analysis for Industrial Equipment,Dalian University of Technology,Dalian 116024,China;Collaborative Innovation Center for Advanced Ship and Deep-Sea Exploration(CISSE),Dalian University of Technology,Dalian 116024,China;Hull Design Section,Dalian Shipbuilding Industry Design and Research Institute,Dalian 116005,China)
出处
《中国造船》
EI
CSCD
北大核心
2021年第1期183-191,共9页
Shipbuilding of China
基金
国家自然科学基金(51979034)
大连市科技创新基金项目(2018J12GX057).
关键词
船体分段
精度控制
最大相关熵
高斯核函数
点云配准
hull block
accuracy control
maximum correntropy
Gaussian kernel function
point clouds registration
