利用LMedS算法與特征值法的點(diǎn)云平面擬合方法
信息技術(shù)與網(wǎng)絡(luò)安全 3期
蘇毅輝，陳志聰，吳麗君
(福州大學(xué) 物理與信息工程學(xué)院，福建 福州350108)
摘要： 針對三維點(diǎn)云數(shù)據(jù)中存在的異常點(diǎn)會(huì)對平面擬合過程產(chǎn)生不利的影響，提出了一種將最小平方中值算法(Least Median of Squares，LMedS)與特征值法相結(jié)合的點(diǎn)云平面擬合新方法。首先，通過LMedS算法進(jìn)行多次迭代確定最佳閾值并剔除點(diǎn)云數(shù)據(jù)中的異常點(diǎn)。然后，采用特征值法對剔除完異常點(diǎn)后的點(diǎn)云數(shù)據(jù)進(jìn)行平面擬合，以獲得更加精確的擬合平面參數(shù)解。最后，分別采用最小二乘法、特征值法、RANSAC+主成分分析法與所提出方法對仿真和實(shí)測點(diǎn)云數(shù)據(jù)進(jìn)行平面擬合計(jì)算。實(shí)驗(yàn)結(jié)果表明，相比于其他方法，該方法的平面擬合精度更高，適用于對含有異常點(diǎn)的點(diǎn)云數(shù)據(jù)進(jìn)行平面擬合，具有較高的魯棒性。
中圖分類號(hào)： TP391
文獻(xiàn)標(biāo)識(shí)碼： A
DOI： 10.19358/j.issn.2096-5133.2022.03.007
引用格式： 蘇毅輝，陳志聰，吳麗君. 利用LMedS算法與特征值法的點(diǎn)云平面擬合方法[J].信息技術(shù)與網(wǎng)絡(luò)安全，2022，41(3)：38-43.
文獻(xiàn)標(biāo)識(shí)碼： A
DOI： 10.19358/j.issn.2096-5133.2022.03.007
引用格式： 蘇毅輝，陳志聰，吳麗君. 利用LMedS算法與特征值法的點(diǎn)云平面擬合方法[J].信息技術(shù)與網(wǎng)絡(luò)安全，2022，41(3)：38-43.
A point cloud plane fitting method using LMedS algorithm and eigenvalue method
Su Yihui，Chen Zhicong，Wu Lijun
(College of Physics and Information Engineering，F(xiàn)uzhou University，F(xiàn)uzhou 350108，China)
Abstract： In order to eliminate the negative effects on the plane fitting process caused by the outliers in 3D point cloud data,a point cloud plane fitting method combining Least Median of Squares(LMedS) algorithm with eigenvalue method was proposed.Firstly, the optimal threshold was determined by multiple iterations of the LMedS, and the outliers in the original point cloud data were removed through LMedS. Then,the eigenvalue method was used to perform plane fitting on the remaining point cloud data in order to obtain more accurate plane parameter solution.The least square method, eigenvalue method, RANSAC+principal component analysis method and the proposed method were used to fit the simulated and measured point cloud data.The experimental results show that the plane fitting accuracy of the proposed method is higher compared with other methods,this method is suitable for fitting the point cloud data with outliers, which has higher robustness.
Key words : point cloud data；outliers；LMedS algorithm；eigenvalue method；plane fitting
0 引言
點(diǎn)云數(shù)據(jù)作為一種能夠完整表達(dá)三維場景信息的重要數(shù)據(jù)，在計(jì)算機(jī)視覺、三維重建、自動(dòng)駕駛等許多領(lǐng)域有廣泛的應(yīng)用[1]。近年來，隨著三維激光掃描技術(shù)的快速發(fā)展，點(diǎn)云數(shù)據(jù)處理現(xiàn)已成為一個(gè)熱門研究話題[2]。點(diǎn)云平面擬合是一種通過建立三維平面模型來和點(diǎn)云數(shù)據(jù)進(jìn)行匹配，使模型和點(diǎn)云數(shù)據(jù)高度吻合并得到最佳模型參數(shù)的數(shù)據(jù)處理方法。三維場景通常是由大量平面組成，比如道路、橋梁、建筑物等都含有平面特征。將三維場景的點(diǎn)云數(shù)據(jù)擬合成平面，再經(jīng)過點(diǎn)云配準(zhǔn)、去噪、簡化后可精確地重建出場景的三維模型，因此點(diǎn)云平面擬合對于三維建模具有極高的研究和應(yīng)用價(jià)值[3-5]。
本文詳細(xì)內(nèi)容請下載：http://ihrv.cn/resource/share/2000004024
作者信息：
蘇毅輝，陳志聰，吳麗君
(福州大學(xué) 物理與信息工程學(xué)院，福建 福州350108)
此內(nèi)容為AET網(wǎng)站原創(chuàng)，未經(jīng)授權(quán)禁止轉(zhuǎn)載。