A 4PCS Coarse Registration Algorithm Based on ISS Feature Points

Yang, Zhongfan; Wang, Xiaogang; Hou, Jin

doi:10.23919/ccc52363.2021.9549486

Cited by 5 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Point cloud registration is a vital component of the LO algorithm, with the objective of computing transformations by minimizing a distance function between adjacent point clouds [25]. Commonly used point cloud registration methods in LO can be primarily categorized into three types: point-based methods, distribution-based methods [26,27], and feature-based methods [28][29][30]. The iterative closest point (ICP) algorithm [31] is perhaps the most widely applied point-based registration method.…”

Section: Point Cloud Registration Methodsmentioning

confidence: 99%

An Iterative Closest Point Method for Lidar Odometry with Fused Semantic Features

Cao,

Liao,

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

Lidar sensors play a pivotal role in a multitude of remote sensing domains, finding extensive applications in various sectors, including robotics, unmanned aerial vehicles (UAVs), autonomous driving, and 3D reconstruction, among others. Their significance and versatility in these areas have made them indispensable tools for a wide range of applications. The accuracy of Lidar odometry (LO), which serves as the front end of SLAM, is crucial. In this paper, we propose a novel iterative closest point (ICP) technique that combines semantic features to improve LO precision. First, the semantic segmentation neural network is used to extract the semantic features from each frame of the point cloud. Then, the obtained semantic features assist in extracting the local geometric features of the point cloud. Also, the residual blocks of the ICP algorithm’s least squares are combined with semantic confidence functions to better predict the exact pose. Compared to LOAM, there is an average improvement of 4 cm in accuracy per 100 m. Experimental results illustrate the superiority of the proposed method and indicate that the fusion of semantic features can robustly improve the precision of LO.

show abstract

Section: Point Cloud Registration Methodsmentioning

confidence: 99%

An Iterative Closest Point Method for Lidar Odometry with Fused Semantic Features

Cao,

Liao,

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…Combinar diferentes modelos e descritores é uma prática comum, como ilustrado por Yang et al (2021) que experimentou a integração do 4PCS com descritores como o SIFT-3D (RISTER et al, 2017), ISS (ZHONG, 2009) e HARRIS-3D (SIPIRAN; BUSTOS, 2011). Esses descritores, inspirados em técnicas de registro de imagens, exploram os autovalores da MVC e Diferenças de Gaussianas (Difference of Gaussians -DoG) em 3D (HARRIS; STEPHENS, 1988), (LOWE, 2004).…”

Section: Registro De Nuvens De Pontos 3d -Estado Da Arteunclassified

Registro Automático de Nuvens de Pontos 3D e Refinamento Global de Poses – Contribuições para o Mapeamento e Localização Simultâneo (SLAM)

Benevides,

Santos,

Pavan

2024

Rev. Bras. Cartogr.

View full text Add to dashboard Cite

O registro de nuvens de pontos 3D e o refinamento global de poses são dois problemas fundamentais ao realizar o Mapeamento e Localização Simultâneos (Simultaneous Localization and Mapping - SLAM) com sensores LIDAR. O registro de nuvens consiste em encontrar transformações de coordenadas que sobrepõem localmente pares de nuvens de pontos, chamadas poses relativas. Para referenciar várias nuvens em uma origem global, várias poses relativas precisam ser compostas multiplicativamente em poses absolutas ao longo da trajetória do sensor, como as poses relativas nunca estão isentas de erros, um problema ainda mais geral surge, a deriva (drift) da trajetória do sensor. Para tratar este problema se utilizam Modelos de Refinamento Global (MRG), que refinam simultaneamente todas as poses de uma trajetória. Neste contexto, propõe-se aqui duas contribuições, a primeira, é um método de registro de pares de nuvens de pontos que integra o Fast Global Registration (FGR) e o Generalized Iterative Closest Point (GICP) em uma abordagem multicaminho e em multiescala. Para isto, cada nuvem de um dataset é registrada nas 3 posteriores, criando um grafo de poses, e cada par é sucessivamente registrado em abordagem coarse-to-fine. A segunda contribuição, se trata de um MRG linear e fechado capaz de refinar todas as poses de um circuito, sem necessidade de iterações ou definição de parâmetros. Para isto, as rotações das poses são mapeadas em quatérnios e interpoladas por meio da técnica Spherical Linear Interpolation (SLERP). Em seguida outra otimização linear baseada no modelo LUM-3D é aplicada. A combinação dos modelos foi testada em dois datasets distintos, um com sete nuvens de pontos obtidas por Laser Scanner Terrestre (LST) e outro com 901 nuvens obtidas por Laser Scaner Móvel (LSM). Em ambos, os modelos foram capazes de reconstruir totalmente os datasets e reduzir significativamente os erros de registro e deriva.

show abstract

“…Xing et al [16] transitioned from the classical 2D scale-invariant feature transform (SIFT) descriptor and extended it to 3D for feature point extraction, enabling the detection of feature points at various scales. Yang et al [17] proposed a coarse point cloud registration method based on ISS feature point descriptors combined with 4PCS and achieved good robustness with respect to noise and density variations. Sun et al [18] proposed an approach to extract feature points by using curvature and fast point feature histograms (FPFHs) for fusion.…”

Section: Introductionmentioning

confidence: 99%

Point Cloud Registration Algorithm Based on Adaptive Neighborhood Eigenvalue Loading Ratio

Liao,

Peng,

Tang

et al. 2024

Applied Sciences

View full text Add to dashboard Cite

Traditional iterative closest point (ICP) registration algorithms are sensitive to initial positions and easily fall into the trap of locally optimal solutions. To address this problem, a point cloud registration algorithm is put forward in this study based on adaptive neighborhood eigenvalue loading ratios. In the algorithm, the resolution of the point cloud is first calculated and used as an adaptive basis to determine the raster widths and radii of spherical neighborhoods in the raster filtering; then, the adaptive raster filtering is implemented to the point cloud for denoising, while the eigenvalue loading ratios of point neighborhoods are calculated to extract and match the contour feature points; subsequently, sample consensus initial alignment (SAC-IA) is used to carry out coarse registration; and finally, a fine registration is delivered with KD-tree-accelerated ICP. The experimental results of this study demonstrate that the feature points extracted with this method are highly representative while consuming only 35.6% of the time consumed by other feature point extraction algorithms. Additionally, in noisy and low-overlap scenarios, the registration error of this method can be controlled at a level of 0.1 mm, with the registration speed improved by 56% on average over that of other algorithms. Taken together, the method in this study cannot only ensure strong robustness in registration but can also deliver high registration accuracy and efficiency.

show abstract

A 4PCS Coarse Registration Algorithm Based on ISS Feature Points

Cited by 5 publications

References 12 publications

An Iterative Closest Point Method for Lidar Odometry with Fused Semantic Features

An Iterative Closest Point Method for Lidar Odometry with Fused Semantic Features

Registro Automático de Nuvens de Pontos 3D e Refinamento Global de Poses – Contribuições para o Mapeamento e Localização Simultâneo (SLAM)

Point Cloud Registration Algorithm Based on Adaptive Neighborhood Eigenvalue Loading Ratio

Contact Info

Product

Resources

About