Rubens Antonio Leite Benevides

Prof. Rubens Antonio Leite Benevides

orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 August 2024 in Sensors
Verified
Journal Article
Open access
PDF file
RBRubens Antonio Leite Benevides
Rubens Antonio Leite Benevides
Daniel Rodrigues Dos SantosDaniel Rodrigues Dos Santos
Daniel Rodrigues Dos Santos
NPNadisson Luis Pavan
Nadisson Luis Pavan
LVLuis Augusto Koenig Veiga
Luis Augusto Koenig Veiga
Global pose refinement is a significant challenge within Simultaneous Localization and Mapping (SLAM) frameworks. For LIDAR-based SLAM systems, pose refinement is integral to correcting drift caused by the successive registration of 3D point clouds collected by the sensor. A divergence between the actual and calculated platform paths characterizes this error. In response to this challenge, we propose a linear, parameter-free model that uses a closed circuit for global trajectory corrections. Our model maps rotations to quaternions and uses Spherical Linear Interpolation (SLERP) for transitions between them. The intervals are established by the constraint set by the Least Squares (LS) method on rotation closure and are proportional to the circuit’s size. Translations are globally adjusted in a distinct linear phase. Additionally, we suggest a coarse-to-fine pairwise registration method, integrating Fast Global Registration and Generalized ICP with multiscale sampling and filtering. The proposed approach is tested on three varied datasets of point clouds, including Mobile Laser Scanners and Terrestrial Laser Scanners. These diverse datasets affirm the model’s effectiveness in 3D pose estimation, with substantial pose differences and efficient pose optimization in larger circuits.

17 October 2022 in Research, Society and Development
Verified
Journal Article
Open access
PDF file
RBRubens Antonio Leite Benevides
Rubens Antonio Leite Benevides
Kalima Mendes PitombeiraKalima Mendes Pitombeira
Kalima Mendes Pitombeira
JCJorge Centeno
Jorge Centeno
PSPaulo Rodrigo Simões
Paulo Rodrigo Simões
O registro de imagens é um problema comum na visão computacional com diversas aplicações que consiste em encontrar a correta transformação entre pares de imagens que se sobrepõem. Neste trabalho objetiva-se apresentar um modelo automático e preciso para registro de imagens utilizando o descritor SIFT e o método de estimação RANSAC adaptado. O registro ocorre através da estimativa da homografia entre os pares de imagens, que utilizam as correspondências pontuais dadas pelo descritor SIFT. As correspondências são classificadas usando um limiar de erro estimado dinamicamente. A análise considera a dispersão do resíduo em vários limites de erro e adota aquele que minimiza a dispersão e a magnitude do erro. O modelo é testado com 8 pares heterogêneos de imagens divididos em dois grupos: 4 pares obtidos com uma câmera profissional e 4 pares obtidos com uma câmera comum. Devido à alta qualidade das imagens do primeiro grupo, poucas iterações do modelo são necessárias para a estimativa da homografia correta. No segundo grupo, o modelo mostrou que é capaz de construir mosaicos entre pares de imagens com sobreposição inferior a 20%, encontrando correspondências exatas entre pares de imagens independentemente do método de aquisição. Além disso, foi capaz de lidar com até 65% de corrupção entre as correspondências, com um tempo total de execução de alguns segundos.

11 January 2024 in Preprints.org
Verified
Preprint
Open access
PDF file
RBRubens Antonio Leite Benevides
Rubens Antonio Leite Benevides
Daniel Rodrigues Dos SantosDaniel Rodrigues Dos Santos
Daniel Rodrigues Dos Santos
NPNadisson Luis Pavan
Nadisson Luis Pavan
Global pose refinement is a significant challenge within Simultaneous Localization and Mapping (SLAM) frameworks. For LIDAR-based SLAM systems, pose refinement is integral to correcting drift caused by the successive registration of 3D point clouds collected by the sensor. A divergence between the actual and calculated platform path characterizes this error. In response to this challenge, we propose a linear, parameter-free model, which uses a closed circuit for global trajectory corrections. Our model maps rotations to quaternions and uses Spherical Linear Interpolation (SLERP) for transitions between them. The intervals are established from the constraint set by the Least Squares (LS) method on rotation closure and are proportional to the circuit's size. Translations are globally adjusted in a distinct linear phase. Additionally, we suggest a coarse-to-fine pairwise registration method, integrating Fast Global Registration and Generalized ICP with multiscale sampling and filtering. The proposed approach is tested on three varied datasets of point clouds, including Mobile Laser Scanners and Terrestrial Laser Scanners. These diverse datasets affirm the model's effectiveness in 3D pose estimation with substantial pose differences and efficient pose optimization in larger circuits.

21 November 2025 in Sensors
Verified
Journal Article
Open access
PDF file
RBRubens Antonio Leite Benevides
Rubens Antonio Leite Benevides
Daniel Rodrigues Dos SantosDaniel Rodrigues Dos Santos
Daniel Rodrigues Dos Santos
LVLuis Augusto Koenig Veiga
Luis Augusto Koenig Veiga
Laser scanning allows for the rapid acquisition of three-dimensional data in the form of 3D point clouds. However, due to the accumulation of errors in the registration of multiple pairs of point clouds along the sensor’s trajectory, the generated 3D reconstructions exhibit drift, which creates global inconsistencies in the scan. To address this error, there are drift correction models that distribute the error along a closed circuit of stations. In this work, we present a model of this nature based on the linear interpolation of dual quaternions. This linear solution simultaneously refines rotations and translations in a closed trajectory without iterative computations or matrix decomposition. Experimental evaluations on eight TLS datasets indicate that the proposed drift correction model provides a robust average error reduction of 26%, with a maximum reduction of 41% in circuits with large drift. This simultaneous solution improves pose accuracy in closed trajectories with theoretical advantages that translate into efficient and fast implementation. Although validated using TLS data, the proposed pose-circuit correction model is sensor-agnostic and can be applied to other 3D mapping systems.

5 March 2024 in Revista Brasileira de Cartografia
Verified
Journal Article
Open access
PDF file
RBRubens Antonio Leite Benevides
Rubens Antonio Leite Benevides
Daniel Rodrigues Dos SantosDaniel Rodrigues Dos Santos
Daniel Rodrigues Dos Santos
NPNadisson Luis Pavan
Nadisson Luis Pavan
3D point cloud registration and global pose refinement are two fundamental problems when performing Simultaneous Localization and Mapping (SLAM) with LIDAR sensors. Cloud registration consists of finding coordinate transformations that locally overlap pairs of point clouds, called relative poses. In order to reference several clouds at a global origin, several relative poses need to be multiplicatively composed into absolute poses along the sensor trajectory, as relative poses are never error-free, an even more general problem arises, the drift of the sensor trajectory. To deal with this, Global Refinement Models (GRM) are used, which simultaneously refine all the poses in a trajectory. In this context, two contributions are proposed here: the first is a method for registering pairs of point clouds that integrates Fast Global Registration (FGR) and Generalized Iterative Closest Point (GICP) in a multipath and multiscale approach. To do this, each cloud in a dataset is registered in the next 3, creating a graph of poses, and each pair is successively registered in a coarse-to-fine approach. The second contribution is a linear, closed MRG capable of refining all the poses in a circuit, without the need for iterations or parameter setting. To do this, the rotations of the poses are mapped onto quaternions and interpolated using the Spherical Linear Interpolation (SLERP) technique. Then another linear optimization based on the LUM-3D model is applied. The combination of models was tested on two different datasets, one with seven point clouds obtained by Laser Scanner Terrestrial (LST) and the other with 901 clouds obtained by Laser Scanner Mobile (LSM). In both cases, the models were able to fully reconstruct the datasets and significantly reduce registration and drift errors.

1 October 2025 in The Photogrammetric Record
Verified
Journal Article
Open access
RBRubens Antonio Leite Benevides
Rubens Antonio Leite Benevides
DSDaniel Rodrigues dos Santos
Daniel Rodrigues dos Santos
LVLuis Augusto Koenig Veiga
Luis Augusto Koenig Veiga
Terrestrial laser scanning (TLS) enables the rapid acquisition of three‐dimensional data in the form of 3D point clouds. However, it presents significant challenges, such as lengthy registration times between point cloud pairs and considerable trajectory drift due to widely spaced stations, which leads to inconsistent 3D reconstruction along the TLS path. This research proposes a pipeline that adapts the fast global registration (FGR) algorithm to efficiently handle TLS‐generated point clouds. The approach includes fine‐tuning FGR parameters and additional preprocessing steps, specifically normal orientation and keypoint extraction. The second contribution introduces a global refinement model (GRM) based on linear interpolation of dual quaternions. This closed‐form solution simultaneously refines rotations and translations in a closed circuit without iterative computations or matrix decomposition. Experimental evaluations on four TLS datasets indicate that the proposed pairwise registration with FGR achieves a 90% success rate across 86 point‐cloud pairs from multiple environments. Moreover, our drift‐correction model reduces closure errors by up to 41% in the dataset circuits, improving pose accuracy in closed trajectories with theoretical advantages that translate into efficient and fast implementation.