本文關(guān)鍵詞： 建筑物 平面提取 層次聚類 平面基元組 點云配準(zhǔn) 隨機采樣一致性(RANSAC)　出處：《武漢大學(xué)學(xué)報(信息科學(xué)版)》2016年12期 　論文類型：期刊論文

【摘要】：三維激光掃描點云在建筑物場景配準(zhǔn)中存在同名特征難以區(qū)分、投票匹配算法復(fù)雜度高等問題。為此,提出了一種基于平面基元組的建筑物場景點云自動配準(zhǔn)方法。平面基元組被定義為具有近似相同法向量的點云面集合。該方法從點云中提取平面基元,根據(jù)平面法向量方向,劃分平面基元組,然后借助基元組搜索同名平面,利用單位四元數(shù)法估計轉(zhuǎn)換參數(shù)。實驗結(jié)果表明,該方法適用于建筑物場景,能夠?qū)崿F(xiàn)點云的自動配準(zhǔn)。
[Abstract]:3D laser scanning point cloud in building scene registration has problems such as hard to distinguish the same name features, high complexity of voting matching algorithm and so on. This paper presents an automatic registration method for building scene point clouds based on planar primitives, which is defined as a set of point cloud surfaces with approximately the same normal vectors. The method extracts plane primitives from point clouds and extracts them according to the direction of plane normal vectors. The transformation parameters are estimated by using the unit quaternion method. The experimental results show that this method is suitable for building scene and can realize the automatic registration of the point cloud by using the primitive group to search the plane with the same name and to estimate the conversion parameters by using the unit quaternion method.
【作者單位】： 中國地質(zhì)大學(xué)(武漢)信息工程學(xué)院;
【基金】：國家自然科學(xué)基金(41471355) 中央高�；究蒲袠I(yè)務(wù)費專項資金(CUGL120267)~~
【分類號】：TU198;TN249
，

本文編號：1536311