Set Of Point Clouds Research Articles

Practical semiautomatic global registration of multiple point clouds based on semidefinite programming

Multiple point clouds registration is a prerequisite step for geometry object observation from multiple viewpoints. Most algorithms depend on pose graph or relative poses, and they make the results ambiguous and inconsistent with the original point clouds. However, the existing algorithms based on original data can almost exclusively be applied to simulated data. In this paper, a practical semiautomatic global registration algorithm is proposed. A objective function is proposed; it utilizes correspondences between point cloud pairs directly to describe this problem, making the algorithm easy to interface with other point cloud matching algorithms. Moreover, it can be effectively approximated by semidefinite programming. Since no automatic algorithm can guarantee the correctness of the correspondences, a manual strategy is proposed to filter the correspondences and find the correlation between point clouds. Based on the correlation, irrelevant point clouds can be removed easily in the point cloud set or divide the set into multiple independent parts. Besides, a postprocessing strategy based on the correlation is proposed for refining the results. The qualitative and quantitative experimental results demonstrate that the approach achieves more accurate and robust performance than previous algorithms.

Robust denoising-oriented point cloud plane fitting method based on a combined model of improved Cook’s distance and WTLS

PurposeIn order to improve the robustness to noise in point cloud plane fitting, a combined model of improved Cook’s distance (ICOOK) and WTLS is proposed by setting a modified Cook’s increment, which could help adaptively remove the noise points that exceeds the threshold.Design/methodology/approachThis paper proposes a robust point cloud plane fitting method based on ICOOK and WTLS to improve the robustness to noise in point cloud fitting. The ICOOK to denoise the initial point cloud was set and verified with experiments. In the meanwhile, weighted total least squares method (WTLS) was adopted to perform plane fitting on the denoised point cloud set to obtain the plane equation.Findings(a) A threshold-adaptive Cook’s distance method is designed, which can automatically match a suitable threshold. (b) The ICOOK is fused with the WTLS method, and the simulation experiments and the actual fitting of the surface of the DD motor are carried out to verify the actual application. (c) The results shows that the plane fitting accuracy and unit weight variance of the algorithm in this paper are substantially enhanced.Originality/valueThe existing point cloud plane fitting methods are not robust to noise, so a robust point cloud plane fitting method based on a combined model of ICOOK and WTLS is proposed. The existing point cloud plane fitting methods are not robust to noise, so a robust point cloud plane fitting method based on a combined model of ICOOK and WTLS is proposed.

The Influence of Flight Direction and Camera Orientation on the Quality Products of UAV-Based SfM-Photogrammetry

Unmanned aerial vehicles (UAVs) can provide valuable spatial information products for many projects across a wide range of applications. One of the major challenges in this discipline is the quality of positioning accuracy of the resulting mapping products in professional photogrammetric projects. This is especially true when using low-cost UAV systems equipped with GNSS receivers for navigation. In this study, the influence of UAV flight direction and camera orientation on positioning accuracy in an urban area on the west bank of the Euphrates river in Iraq was investigated. Positioning accuracy was tested in this study with different flight directions and camera orientation settings using a UAV autopilot app (Pix4Dcapture software (Ver. 4.11.0)). The different combinations of these two main parameters (camera orientation and flight direction) resulted in 11 different flight cases for which individual planimetric and vertical accuracies were evaluated. Eleven flight sets of dense point clouds, DEMs, and ortho-imagery were created in this way to compare the achieved positional accuracies. One set was created using the direct georeferencing method (without using GCPs), while the other ten sets were created using the indirect georeferencing approach based on ground truth measurements of five artificially created GCPs. Positional accuracy was found to vary depending on the user-defined flight plan settings, despite an approximately constant flight altitude. However, it was found that the horizontal accuracy achieved was better than the vertical accuracy for all flight sets. This study revealed that combining multiple sets of images with different flight directions and camera orientations can significantly improve the overall positional accuracy to reach several centimeters.

Segmentation and Stratification Methods of Field Maize Terrestrial LiDAR Point Cloud

Three-dimensional (3D) laser point cloud technology is an important research method in the field of agricultural remote sensing research. The collection and processing technology of terrestrial light detection and ranging (LiDAR) point cloud of crops has greatly promoted the integration of agricultural informatization and intelligence. In a smart farmland based on 3D modern agriculture, the manager can efficiently and conveniently achieve the growth status of crops through the point cloud collection system and processing model integrated in the smart agricultural system. To this end, we took field maize as the research object in this study and processed four sets of field maize point clouds, named Maize-01, Maize-02, Maize-03, and Maize-04, respectively. In this research, we established a field individual maize segmentation model with the density-based clustering algorithm (DBSCAN) as the core, and four groups of field maize were used as research objects. Among them, the value of the overall accuracy (OA) index, which was used to evaluate the comprehensive performance of the model, were 0.98, 0.97, 0.95, and 0.94. Secondly, the multi-condition identification method was used to separate different maize organ point clouds from the individual maize point cloud. In addition, the organ stratification model of field maize was established. In this organ stratification study, we take Maize-04 as the research object and obtained the recognition accuracy rates of four maize organs: tassel, stalk, ear, and leaf at 96.55%, 100%, 100%, and 99.12%, respectively. We also finely segmented the leaf organ obtained from the above-mentioned maize organ stratification model into each leaf individual again. We verified the accuracy of the leaf segmentation method with the leaf length as the representative. In the linear analysis of predicted values of leaf length, R2 was 0.73, RMSE was 0.12 m, and MAE was 0.07 m. In this study, we examined the segmentation of individual crop fields and established 3D information interpretations for crops in the field as well as for crop organs. Results visualized the real scene of the field, which is conducive to analyzing the response mechanism of crop growth and development to various complex environmental factors.

Extraction of Submarine Gas Plume Based on Multibeam Water Column Point Cloud Model

The gas plume is a direct manifestation of sea cold seep and one of the most significant symbol indicators of the presence of gas hydrate reservoirs. The multibeam water column (MWC) data can be used to extract and identify the gas plume efficiently and accurately. The current research methods mostly start from the perspective of image theory, which cannot identify the three-dimensional (3D) spatial structure features of gas plumes, reducing the efficiency and accuracy of detection. Therefore, this paper proposes a method for identifying and extracting the gas plume based on an MWC point cloud model, which calculates the spatially resolved homing of MWC data and constructs a 3D point cloud model of MWC containing acoustic reflection intensity information. It first performs noise suppression of the 3D point cloud of the MWC based on the symmetric subtraction and Otsu algorithm by leveraging the noise distribution of the MWC and the reflection intensity characteristics of the gas plume. Then, it extracts the point cloud clusters containing the gas plume based on Density-Based Spatial Clustering of Applications with Noise (DBSCAN) according to the density difference between the gas plume point cloud and the background MWC point cloud and next identifies the point cloud clusters by feature matching based on fast point feature histograms (FPFHs). Finally, it extracts the gas plume point cloud set in the MWC. As evidenced by the MWC data collected from gas hydrate enrichment zones in the Gulf of Mexico, the location of gas plume extracted by this method is highly consistent with that of gas leakage points measured during the cruise. Using this method, we obtained the point cloud data set of gas plume for the first time and accurately characterized the 3D spatial morphology of the subsea gas plume, providing technical support for gas hydrate exploration, subsea gas seepage area delineation, and subsea seepage gas flux estimation.

Structure-Aware Denoising for Real-world Noisy Point Clouds with Complex Structures

Point cloud denoising is a crucial and fundamental step in geometry processing, which has achieved significant progress in the last two decades. Denoising real-world noisy point clouds is a very challenging problem since it is hard to describe the complex real-world noise by simple distributions such as Gaussian distribution. Furthermore, existing methods may suffer from performance degradation when dealing with real-world noisy point clouds with complex structures, which contain not only sharp features (sharp edges, sharp corners, etc.) but also smooth features, fine features, etc. To solve the above-mentioned problems, we propose a novel structure-aware denoising approach by exploiting the prior information in both external clean point clouds and the given noisy point cloud. We first group nonlocal self-similarity (NSS) patches from a set of external clean point clouds. Then, we employ the Gaussian Mixture Model (GMM) learning algorithm to learn external NSS priors over patch groups. Next, the internal priors are learned from the given noisy point cloud in the same way to refine the prior model. We integrate both the learned external and internal priors into a set of orthogonal dictionaries to efficiently estimate point normals. Finally, we propose a feature-aware point updating method through adaptive neighborhood selection to reposition points to match the estimated normals. Extensive experiments show that our approach achieves favorable comprehensive performance compared with many popular or state-of-the-art methods in terms of both objective and visual perception. The source code can be found at https://zhiyongsu.github.io.

LiODOM: Adaptive local mapping for robust LiDAR-only odometry

In the last decades, Light Detection And Ranging (LiDAR) technology has been extensively explored as a robust alternative for self-localization and mapping. These approaches typically state ego-motion estimation as a non-linear optimization problem dependent on the correspondences established between the current point cloud and a map, whatever its scope, local or global. This paper proposes LiODOM, a novel LiDAR-only ODOmetry and Mapping approach for pose estimation and map-building, based on minimizing a loss function derived from a set of weighted point-to-line correspondences with a local map abstracted from the set of available point clouds. Furthermore, this work places a particular emphasis on map representation given its relevance for quick data association. To efficiently represent the environment, we propose a data structure that combined with a hashing scheme allows for fast access to any section of the map. LiODOM is validated by means of a set of experiments on public datasets, for which it compares favourably against other solutions. Its performance on-board an aerial platform is also reported.

Registration Strategy of Handheld Scanner (HS) and Terrestrial Laser Scanner Integration for Building Utility Mapping

Currently, the need for a 3D model that represents the existing condition of a building is needed, especially for building management. While performed the 3D mapping of building utilities such as pipelines in buildings, the Terrestrial Laser Scanner (TLS) technology couldn’t cover the object in the ceiling, so Handheld Scanner (HS) was used to complete the scanned objects. The purpose of this article is conducting the process of mapping between HS and TLS for building pipelines. The integration between two different sensor resolutions and sensor scan coverage requires a separate strategy to generate 3D models for pipeline objects in building. The use of specific target and 3D transformation method could do the registration between two set of point clouds which was obtained by two different technologies. The target object, such as spherical object which is usually used as a tie point between scan results in the registration process, cannot be recorded by HS. The specific targets were pasted on the pipelines in order to be captured by the HS. The specific targets become tie point in the registration process. Meanwhile, the registration process using the Iterative Closest Point (ICP) algorithm cannot be carried out because the two scan results do not meet the overlap percentage standard, so that the 3D transformation would be used in registration process. The result is that the registration accuracy is 0.062 m.

Automatic Extrinsic Calibration of 3D LIDAR and Multi-Cameras Based on Graph Optimization.

In recent years, multi-sensor fusion technology has made enormous progress in 3D reconstruction, surveying and mapping, autonomous driving, and other related fields, and extrinsic calibration is a necessary condition for multi-sensor fusion applications. This paper proposes a 3D LIDAR-to-camera automatic calibration framework based on graph optimization. The system can automatically identify the position of the pattern and build a set of virtual feature point clouds, and can simultaneously complete the calibration of the LIDAR and multiple cameras. To test this framework, a multi-sensor system is formed using a mobile robot equipped with LIDAR, monocular and binocular cameras, and the pairwise calibration of LIDAR with two cameras is evaluated quantitatively and qualitatively. The results show that this method can produce more accurate calibration results than the state-of-the-art method. The average error on the camera normalization plane is 0.161 mm, which outperforms existing calibration methods. Due to the introduction of graph optimization, the original point cloud is also optimized while optimizing the external parameters between the sensors, which can effectively correct the errors caused during data collection, so it is also robust to bad data.

Active Stereo Vision System with Rotated Structured Light Patterns and Two-Step Denoising Process for Improved Spatial Resolution

Three-dimensional (3D) scanning systems are widely used nowadays for multiple applications. If the object is too large, it is a challenge to provide the detailed characteristics of the scanned object. Improving the spatial resolution of the scanning process is essential in improving the quality of the reconstruction results. Accordingly, this study has proposed an active stereo vision system for improving the spatial resolution. It is well known that the reconstruction quality of stereo vision systems is severely impaired for objects with limited features on their surfaces, thereby, in the present study, an active stereo vision system is proposed in which stripe patterns are projected onto the object surface with various rotation angles in order to facilitate a more accurate and complete detection of the matching points. Notably, the point clouds obtained by the projected patterns at each rotation angle are formed within the same world coordinate system. As a result, the final reconstructed 3D object can be obtained by merging sets of point clouds directly without the need for a prior registration process. In practice, the quality of the reconstruction results obtained by stereo vision systems is commonly degraded by the effects of noise. Therefore, this study also proposes a two-step (multi-decoding pattern and Epipolar line) method for removing the incorrect matching points and improving the reconstruction results accordingly. The performance of the proposed reconstruction system is evaluated experimentally and compared with that of a general active stereo vision system. The experimental results show that the active stereo vision system presented in the present study provides a low-cost, convenient and effective approach for performing the reconstruction of 3D objects with high spatial resolution.

Person Reidentification Based on Automotive Radar Point Clouds

Person reidentification (ReID) systems play a key role in intelligent visual surveillance systems and have widespread applications, for example, in public security. Usually, person ReID systems can identify a person with cameras. In this article, we focus on the relatively unexplored area of using low-cost automotive radar for the person ReID problem. Unlike the radar-based person identification, person ReID has some characteristics, such as the uncooperative scenes and the long-term robustness. Therefore, we design a new deep learning network to extract spatiotemporal information from 4-D radar point clouds. We also build a data set of radar point clouds collected from the real-world person ReID scenarios. The evaluation result shows that our method achieves 91% CMC-1 accuracy on the ReID task. Besides, for the person identification task, our method also achieves accuracies of 98% and 91% for 15 and 40 individuals, respectively. In addition, we discuss the potential of using radar for person ReID problems and intuitively explain the new method’s performance. Finally, we analyze the robustness and the influence of different parameters on the method and the contributions of different modules to the network model. The results of our experiment indicate that radar-based ReID not only preserves privacy but also outperforms camera-based ReID in some cases, such as in low-light environments or with substantial clothing changes.

PU-Dense: Sparse Tensor-Based Point Cloud Geometry Upsampling.

Due to the increased popularity of augmented and virtual reality experiences, the interest in capturing high-resolution real-world point clouds has never been higher. Loss of details and irregularities in point cloud geometry can occur during the capturing, processing, and compression pipeline. It is essential to address these challenges by being able to upsample a low Level-of-Detail (LoD) point cloud into a high LoD point cloud. Current upsampling methods suffer from several weaknesses in handling point cloud upsampling, especially in dense real-world photo-realistic point clouds. In this paper, we present a novel geometry upsampling technique, PU-Dense, which can process a diverse set of point clouds including synthetic mesh-based point clouds, real-world high-resolution point clouds, real-world indoor LiDAR scanned objects, as well as outdoor dynamically acquired LiDAR-based point clouds. PU-Dense employs a 3D multiscale architecture using sparse convolutional networks that hierarchically reconstruct an upsampled point cloud geometry via progressive rescaling and multiscale feature extraction. The framework employs a UNet type architecture that downscales the point cloud to a bottleneck and then upscales it to a higher level-of-detail (LoD) point cloud. PU-Dense introduces a novel Feature Extraction Unit that incorporates multiscale spatial learning by employing filters at multiple sampling rates and receptive fields. The architecture is memory efficient and is driven by a binary voxel occupancy classification loss that allows it to process high-resolution dense point clouds with millions of points during inference time. Qualitative and quantitative experimental results show that our method significantly outperforms the state-of-the-art approaches by a large margin while having much lower inference time complexity. We further test our dataset on high-resolution photo-realistic datasets. In addition, our method can handle noisy data well. We further show that our approach is memory efficient compared to the state-of-the-art methods.

Ghost Point Diffusion Maps for Solving Elliptic PDEs on Manifolds with Classical Boundary Conditions

In this paper, we extend the class of kernel methods, the so‐called diffusion maps (DM) and its local kernel variants to approximate second‐order differential operators defined on smooth manifolds with boundaries that naturally arise in elliptic PDE models. To achieve this goal, we introduce the ghost point diffusion maps (GPDM) estimator on an extended manifold, identified by the set of point clouds on the unknown original manifold together with a set of ghost points, specified along the estimated tangential direction at the sampled points on the boundary. The resulting GPDM estimator restricts the standard DM matrix to a set of extrapolation equations that estimates the function values at the ghost points. This adjustment is analogous to the classical ghost point method in a finite‐difference scheme for solving PDEs on flat domains. As opposed to the classical DM, which diverges near the boundary, the proposed GPDM estimator converges pointwise even near the boundary. Applying the consistent GPDM estimator to solve well‐posed elliptic PDEs with classical boundary conditions (Dirichlet, Neumann, and Robin), we establish the convergence of the approximate solution under appropriate smoothness assumptions. We numerically validate the proposed mesh‐free PDE solver on various problems defined on simple submanifolds embedded in Euclidean spaces as well as on an unknown manifold. Numerically, we also found that the GPDM is more accurate compared to DM in solving elliptic eigenvalue problems on bounded smooth manifolds. © 2021 Wiley Periodicals LLC.

Experimental Study on 3D Measurement Accuracy Detection of Low Altitude UAV for Repeated Observation of an Invariant Surface

Low-altitude unmanned aerial vehicle (UAV) photogrammetry combined with structure-from-motion (SFM) algorithms is the latest technological approach to imaging 3D stereo constructions. At present, derivative products have been widely used in landslide monitoring, landscape evolution, glacier movement, volume measurement, and landscape change detection. However, there is still a lack of research into the accuracy of 3D data positioning based on the structure-from-motion of unmanned aerial vehicle (UAV-SFM) technology, itself, which can affect the measurable effectiveness of the results in further applications of this technological approach. In this paper, validation work was carried out for the DJI Phantom 4 RTK UAV, for earth observation data related to 3D positioning accuracy. First, a test plot with a relatively stable surface was selected for repeated flight imaging observations. Specifically, three repeated flights were performed on the test plot to obtain three sorties of images; the structure from motion and multi-view stereo (SFM-MVS) key technology was used to process and construct a 3D scene model, and based on this model the digital surface model (DSM) and digital orthophoto map (DOM) data of the same plot with repeated observations were obtained. In order to check the level of 3D measurement accuracy of the UAV technology itself, a window selection-based method was used to sample the point cloud set data from the three-sortie repeat observation 3D model. The DSM and DOM data obtained from three repeated flights over the surface invariant test plots were used to calculate the repeat observation 3D point errors, taking into account the general methodology of redundant observation error analysis for topographic surveys. At the same time, to further analyze the limits of the UAV measurement technique, possible under equivalent observation conditions with the same processing environment, a difference model (DOD) was constructed for the DSM data from three sorties, to deepen the overall characterization of the differences between the DSMs obtained from repeated observations. The results of the experimental study concluded that both the analysis of the 3D point set measurements based on window sampling and the accuracy evaluation using the difference model were generally able to achieve a centimeter level of planimetric accuracy and vertical accuracy. In addition, the accuracy of the surface-stabilized hardened ground was better, overall, than the accuracy of the non-hardened ground. The results of this paper not only probe the measurement limits of this type of UAV, but also provide a quantitative reference for the accurate control and setting of an acquisition scheme of the UAV-based SfM-MVS method for geomorphological data acquisition and 3D reconstruction.

Laser scanner and drone photogrammetry: A statistical comparison between 3-dimensional models and its impacts on outdoor crime scene registration

This work evaluated the accuracy of 3D models generated by a DJI Mavic Pro drone with 3DF Zephyr software photogrammetry. The models were compared to models generated by a Trimble X7 laser scanner. The tests were performed in the outdoor area of a vehicle parking inbound to simulate the characteristics of a crime scene. Ground control points (GCPs) were distributed in ten positions within the surroundings. In manual flight, the drone performed nadiral photographs from one side to the other side and with an elliptical 45° center pointed. Three altitudes where tested: 10 m, 20 m and 40 m. The Trimble X7 laser scanner performed six scans and generated one set of point clouds. Drone photogrammetry returned eligible data for distances of 20 m and 40 m with errors of ~0.25 mm. To increase the overlay in the photogrammetry procedure, all photographs from distances of 10–40 m were processed, returning an error of ~0.53 mm. The results of the measured distances, which were manually picked from the GCPs, from the 3D-scanned model and photogrammetric 3D models were then statistically analyzed. The Trimble X7 laser scanner showed an average error of 3 cm, which was approximately equivalent to the results obtained with all images or when using a known scale value for the drone photographs, presenting no significant differences among the evaluated methods.

Predicting high-fidelity human body models from impaired point clouds

Accurate 3-D models of human subjects are widely used in domains such as fashion design, non-contact body biometrics, computer animation, gaming, AR/VR, to cite a few. For these kinds of applications, a high-fidelity human body mesh in a canonical posture (e.g. pose or pose) is necessary. This paper proposes a deep learning approach to jointly reconstruct a clean, watertight body mesh and to normalize the posture of the human body model starting from an input set of impaired body point clouds. The proposed method, dubbed Impaired-to-High-fidelity human body network (I2H) is, to the best of our knowledge, the first deep learning approach in the literature that addresses these problems. The proposed method follows an Encoder-Decoder design. The Encoder directly takes the impaired point clouds (e.g. containing noise, occlusions and misalignments) as input without making any structural assumptions about the input. The Decoder interprets the latent feature and produces a high-fidelity T-pose body mesh. We compare the proposed approach against existing state-of-the-art methods through various experiments and show that our method achieves the best performance on both synthetic and scanned datasets for 3D human mesh reconstruction.

A Point Cloud Fusion Method for Space Target 3D Laser Point Cloud and Visible Light Image Reconstruction Point Cloud

In this paper, a high-precision spatial target lidar for 3D reconstruction of point cloud and visible light image 3D reconstruction point cloud fusion method is proposed. This fusion method uses the solved 3D reconstruction to invert the camera pose and 3D point cloud model centroid position optimization initial value selection. The registration accuracy and registration efficiency of the ICP algorithm are improved. At the same time, according to the characteristics of the two sets of point clouds, the Euclidean distance threshold is used to delete the noise points of the 3D point cloud edge, and the high-precision 3D reconstruction point with the scale information fusion is obtained. cloud. The simulation experiment of the spatial target simulation model shows that the fusion method can effectively improve the point cloud density, fill the reconstruction vulnerability, and improve the point cloud accuracy of the spatial target 3D reconstruction.

Fast 3D laser scanning of highly reflective surfaces based on a dual-camera system

Existing laser scanning approaches for highly reflective surfaces generally require multiple scans to cope with specular reflection. This paper proposes a dual-camera, line-laser scanning strategy to reconstruct them efficiently, which only needs a single scanning path. To bypass complicated point cloud processing algorithms for outlier elimination, a feasible image processing scheme was designed to delete the disturbed stripe regions. By merging two sets of point clouds computed from residual stripes on a common world frame, these removed noise patches can be compensated by the opposite camera. Therefore, the data fusion result can reconstruct the 3D geometry of targets completely. Experimental results demonstrate good performance with the ability to handle highly reflective surfaces. The scanning period is significantly shortened and scanning accuracy can meet the industrial requirement.

Geometry Guided Network for Point Cloud Registration

Point cloud registration is a well-known way to align two different point clouds via a rigid transform estimation in robotics and computer vision applications. In particular, deep learning-based methods have recently attempted to extract highly distinguishable point features for each point. However, it is still challenging to learn highly distinctive point features in different locations but in a similar shape, bringing performance limitations. To address this issue, we propose Geometry Guided Network, namely G$^2$ Net, for point cloud registration with spherical positional encoding method and unsupervised geometry consistency loss. Combined with self-attention, the positional encoding learns the globally unique features by assigning global geometric positional information into irregular 3D points. The uniqueness of each feature is further strengthened with geometric consistency loss across two different point cloud sets. We demonstrate that G$^2$ Net outperforms current state-of-the-arts models in point cloud registration tasks with both full and partial registration experiments for the ModelNet40 and Augmented ICL-NUIM datasets. Various visualizations on learned features are provided for demonstrating the global shape-awareness of our methodology.

A Strong-Form Off-Lattice Boltzmann Method for Irregular Point Clouds

Radial basis function generated finite differences (RBF-FD) represent the latest discretization approach for solving partial differential equations. Their benefits include high geometric flexibility, simple implementation, and opportunity for large-scale parallel computing. Compared to other meshfree methods, typically based upon moving least squares (MLS), the RBF-FD method is able to recover a high order of algebraic accuracy while remaining better conditioned. These features make RBF-FD a promising candidate for kinetic-based fluid simulations such as lattice Boltzmann methods (LB). Pursuant to this approach, we propose a characteristic-based off-lattice Boltzmann method (OLBM) using the strong form of the discrete Boltzmann equation and radial basis function generated finite differences (RBF-FD) for the approximation of spatial derivatives. Decoupling the discretizations of momentum and space enables the use of irregular point cloud, local refinement, and various symmetric velocity sets with higher order isotropy. The accuracy and computational efficiency of the proposed method are studied using the test cases of Taylor–Green vortex flow, lid-driven cavity, and periodic flow over a square array of cylinders. For scattered grids, we find the polyharmonic spline + poly RBF-FD method provides better accuracy compared to MLS. For Cartesian node layouts, the results are the opposite, with MLS offering better accuracy. Altogether, our results suggest that the RBF-FD paradigm can be applied successfully also for kinetic-based fluid simulation with lattice Boltzmann methods.