摘要
点云配准是三维建模工作中至关重要的一环。传统的正态分布变换(NDT)和迭代最近点(ICP)结合的配准算法对点云初始位姿要求严格且精度不足。为提高接触网腕臂三维建模的效率与准确度,提出了一种融合Harris特征与NDT-ICP的点云配准算法。该方法利用Harris特征点的尺度不变性和旋转不变性来优化点云数据结构并加快配准速度,并通过计算关键点间的距离信息,基于位置相似性建立关键点对对应关系,进一步提升配准精度。与传统NDT-ICP算法相比,新方法的配准精度提高了近99.98%,配准效率提高了近53.79%。
To address the problem of strict initial pose requirements and low accuracy in traditional NDT-ICP algorithms,a novel registration algorithm combining Harris features and NDT-ICP is proposed.This method utilizes the scale and rotation invariance of Harris feature points to optimize the point cloud data structure and accelerate registration speed.Moreover,calculating the distance information between key points and establishing the corresponding relationship of key point pairs based on positional similarity further improves the registration accuracy.Experimental results show that compared with the traditional NDT-ICP algorithm,the proposed method improves registration accuracy by nearly 99.98%and registration efficiency by nearly 53.79%.
作者
张志春
王志坤
陈祎琳
韦民洋
ZHANG Zhichun;WANG Zhikun;CHEN Yilin;WEI Minyang(Shaanxi Railway Logistics Group Co.Ltd,Xi′an,Shaanxi 710065,China;Shaanxi Jing Shen Railway Co.Ltd,Yulin,Shaanxi 719000,China;Electrical Engineering College,Southwest Jiaotong University,Chengdu,Sichuan 611756,China)
出处
《计量学报》
北大核心
2025年第9期1348-1355,共8页
Acta Metrologica Sinica
基金
国家自然科学基金面上项目(62173279)
四川省科技厅中央引导地方项目(2024ZYD0018)。
