DOPNet: Achieving Accurate and Efficient Point Cloud Registration Based on Deep Learning and Multi-Level Features

When measuring surface deformation, because the overlap of point clouds before and after deformation is small and the accuracy of the initial value of point cloud registration cannot be guaranteed, traditional point cloud registration methods cannot be applied. In order to solve this problem, a complete solution is proposed, first, by fixing at least three cones to the target. Then, through cone vertices, initial values of the transformation matrix can be calculated. On the basis of this, the point cloud registration can be performed accurately through the iterative closest point (ICP) algorithm using the neighboring point clouds of cone vertices. To improve the automation of this solution, an accurate and automatic point cloud registration method based on biological vision is proposed. First, the three-dimensional (3D) coordinates of cone vertices are obtained through multi-view observation, feature detection, data fusion, and shape fitting. In shape fitting, a closed-form solution of cone vertices is derived on the basis of the quadratic form. Second, a random strategy is designed to calculate the initial values of the transformation matrix between two point clouds. Then, combined with ICP, point cloud registration is realized automatically and precisely. The simulation results showed that, when the intensity of Gaussian noise ranged from 0 to 1 mr (where mr denotes the average mesh resolution of the models), the rotation and translation errors of point cloud registration were less than 0.1° and 1 mr, respectively. Lastly, a camera-projector system to dynamically measure the surface deformation during ablation tests in an arc-heated wind tunnel was developed, and the experimental results showed that the measuring precision for surface deformation exceeded 0.05 mm when surface deformation was smaller than 4 mm.

Point cloud models are prevalently utilized in the architectural and civil engineering sectors. The registration of point clouds can invariably introduce registration errors, adversely impacting the accuracy of point cloud models. While the domain of computer vision has delved profoundly into point cloud registration, limited research in the construction domain has explored these registration algorithms in the built environment, despite their inception in the field of computer vision. The primary objective of this study is to investigate the impact of mainstream point cloud registration algorithms—originally introduced in the computer vision domain—on point cloud models, specifically within the context of bridge engineering as a category of civil engineering data. Concurrently, this study examines the influence of noise removal on varying point cloud registration algorithms. Our research quantifies potential variables for registration quality based on two metrics: registration error (RE) and time consumption (TC). Statistical methods were employed for significance analysis and value engineering assessment. The experimental outcomes indicate that the GRICP algorithm exhibits the highest precision, with RE values of 3.02 mm and 2.79 mm under non-noise removal and noise removal conditions, respectively. The most efficient algorithm is PLICP, yielding TC values of 3.86 min and 2.70 min under the aforementioned conditions. The algorithm with the optimal cost-benefit ratio is CICP, presenting value scores of 3.57 and 4.26 for non-noise removal and noise removal conditions, respectively. Under noise removal conditions, a majority of point cloud algorithms witnessed a notable enhancement in registration accuracy and a decrease in time consumption. Specifically, the POICP algorithm experienced a 32% reduction in RE and a 34% decline in TC after noise removal. Similarly, PLICP observed a 34% and 30% reduction in RE and TC, respectively. KICP showcased a decline of 23% in RE and 28% in TC, CICP manifested a 27% and 31% drop in RE and TC, respectively, GRICP observed an 8% reduction in RE and a 40% decline in TC, and for FGRICP, RE and TC decreased by 8% and 52%, respectively, subsequent to noise removal.

Point cloud registration from laser scanning data is a technique to establish the mapping relationship between source and target point clouds, which has been widely used in automatic 3D reconstruction, pose estimation, localization, and navigation. While algorithms like Super4PCS and MSSF-4PCS can achieve registration without initial poses, they are relatively slow, less accurate, and require iterations. To address these issues, we propose a 3D point cloud registration algorithm based on interval segmentation and multi-dimensional feature. Firstly, the source and target point clouds are segmented internally and the point cloud curvature is designed to narrow down the search range for the registration between the segmented point clouds. Secondly, the corresponding four-point sets in the segmented areas of the source and target point clouds are determined using affine invariance constraints. Finally, a multi-dimensional feature vector based on curvature features and fast point feature histogram is established to determine the unique corresponding four-point set pairs, and the rigid body transformation matrix is solved accordingly. Our algorithm is tested on publicly available 3D point cloud data models Bunny, Dino, Dragon, and Horse from Stanford University. Results showed that our algorithm improved registration accuracy by 24.39% and registration efficiency by 46.21% compared to the MSSF-4PCS point cloud registration algorithm. Multiple sets of experimental results confirmed the effectiveness of our algorithm. The proposed 3D point cloud registration is proved to be fast with high accuracy, which can be utilized for automatic segmentation, reconstruction, and modelling from Laser Scanning Data.

Fast, accurate registration between image space and patient space is the key point to surgical navigation system. In this paper, we research and implement an face matching method based on biometri. Against to the smooth, textured feature of human face, we built a proactive structured light vision system, to get patient's facial point cloud, proposed and implemented a rough face point cloud registration method based on biometric feature of human face. By identify the nose and eyes in the two-dimensional image space by Active Shape Model (ASM) method which based on statistical information, and identify the corresponding nose and eye by the principal component analysis of the local neighborhood in the three-dimensional space, thereby automatically get an initial registration matrix for those two point cloud. Then accurate complete point cloud registration, based on Iterative Closest Point (ICP) algorithm, achieving space registration.

Although Point Clouds Registration is a very well studied problem, with many different solutions, most of the approaches in the literature aims at aligning two dense point clouds. Instead, we tackle the problem of aligning a dense point cloud with a sparse one: a problem that has to be solved, for example, to merge maps produced by different sensors, such as a vision-based sensor and laser scanner or two different laser-based sensors. The most used approach to point clouds registration, Iterative Closest Point (ICP), is also applicable to this sub-problem. We propose an improvement over the standard ICP data association policy and we called it Probabilistic Data Association. It was derived applying statistical inference techniques on a fully probabilistic model. In our proposal, each point in the source point cloud is associated with a set of points in the target point cloud; each association is then weighted so that the weights form a probability distribution. The result is an algorithm similar to ICP but more robust w.r.t. noise and outliers. While we designed our approach to deal with the problem of dense-sparse registration, it can be successfully applied also to standard point clouds registration.

3D reconstruction has been widely applied in medical images, industrial inspection, self-driving cars, and indoor modeling. The 3D model is built by the steps of data collection, point cloud registration, surface reconstruction, and texture mapping. In the process of data collection, due to the limited visibility of the scanning system, the scanner needs to scan multiple angles and then splice the data to obtain a complete point cloud model. The point clouds from different angles must be merged into a unified coordinate system, which is known as point cloud registration. The result of point cloud registration can directly affect the accuracy of the point cloud model; thus, point cloud registration is a key step in the construction of the point cloud model. The ICP (Iterative Closest Points) algorithm is the most known technique of the point cloud registration. The variational ICP problem can be solved not only by deterministic but also by stochastic methods. One of them is Grey Wolf Optimizer (GWO) algorithm. Recently, GWO has been applied to rough point clouds alignment. In the proposed paper, we apply the GWO approach to the realization of the point-to-point ICP algorithms. Computer simulation results are presented to illustrate the performance of the proposed algorithm.

Point cloud registration is widely used in robot grasping, aerial mapping, 3D reconstruction and other fields. Aiming at the problems of large pose deviation of point clouds in photographic point clouds, large noise and dense number of point clouds, it is difficult for the traditional iterative closest point (ICP) algorithm to achieve automatic registration. In order to solve the above problems, an improved precision point cloud registration method is proposed. The method includes two stages: coarse registration and fine registration, in the coarse registration stage, the point cloud is simplified by the Approximate Intrinsic Voxel Structure (AIVS) algorithm, and the simplified point cloud is initially registered by the improved RANSAC algorithm. Then, on the basis of the initial registration, a fast and accurate algorithm (FR-ICP) is used for precise registration. This method solves the problem that ICP is sensitive to initial pose. Experiments based on the self-constructed auto parts point cloud dataset show that compared with the ICP algorithm, the registration accuracy and time of the proposed method are improved by 95.8% and 67.8%, respectively.

Laser scanners provide a three-dimensional sampled representation of the surfaces of objects with generally a very large number of points. The spatial resolution of the data is much higher than that of conventional surveying methods. Since laser scanners have a limited field of view, it is necessary to collect data from several locations in order to obtain a complete representation of an object. These data must be transformed into a common coordinate system. This procedure is called the registration of point clouds. In terms of input data, registration methods can be classified into two categories: one is the registration of two point clouds from different scanner locations, so-called pair-wise registration (Rusinkiewicz & Levoy, 2001), and the other is simultaneous registration of multiple point clouds (Pulli, 1999; Williams & Bennamoun, 2001). However, the global registration of multiple scans is more difficult because of the large nonlinear search space and the huge number of point clouds involved. Commercial software typically uses separately scanned markers that can be automatically identified as corresponding points. Akca (2003) uses the special targets attached onto the object(s) as landmarks and their 3-D coordinates are measured with a theodolite in a ground coordinate system before the scanning process. Radiometric and geometric information (shape, size, and planarity) are used to automatically find these targets in point clouds by using cross-correlation, the dimension test and the planarity test. According to the automatic registration problems, several efforts have been made to avoid the use of artificial markers. One of the most popular methods is the Iterative Closest Point (ICP) algorithm developed by Besl & McKay(1992) and Chen & Medioni (1992). ICP operates two point clouds and an estimate of the aligning rigid body transform. It then iteratively refines the transform by alternating the steps of choosing corresponding points across the point clouds, and finding the best rotation and translation that minimizes an error metric based on the distance between the corresponding points. One key to this method is to have a good priori alignment. That means, for partially unorganized and overlapping points, if there is lack of good initial alignment, many other ICP variant don’t work well because it becomes very hard to find corresponding points between the point clouds. Also, although a considerable amount of work on registration of point clouds from laser scanners, it is difficult to understand convergence behavior related to different starting conditions, and error metrics. Many experiment showed that the rate of convergence of ICP heavily relies on the choice of the corresponding point-pairs, and the distance function.

RoCNet++: Triangle-based descriptor for accurate and robust point cloud registration

Laser three-dimensional (3D) image is a novel non-cooperative target 3D image acquisition technology, and the improvements in detection capability and imaging accuracy of the system are critically dependent on efficient echo-signal processing technique and 3D reconstruction method. The registration process is an essential step in array 3D imaging laser point cloud data processing. Registration of point clouds is an effective method that solves the problem caused by the target self-occlusion in the laser 3D imaging system. The accurate registration result will help provide better support for subsequent applications, such as object reconstruction and target recognition. In this study, a set of thresholds in the iterative closest point (ICP) algorithm is analysed on the basis of the characteristics of the laser array 3D imaging system and is combined with the range error and visual lateral resolution of the system, which are both important parameters in the imaging system. To improve the accuracy and speed of registration, the stop threshold of the iterative algorithm and the corresponding point-distance threshold in the algorithm are established in a novel way based on the range error and visual lateral resolution of the system. This forms the foundation, based on which an adaptive threshold ICP algorithm is proposed. The principal idea of the algorithm is to improve the threshold set that has a considerable effect on the accuracy and speed of registration. At first, the characteristics of the imaging point clouds of the laser array 3D imaging system are analysed in the algorithm. Based on this analysis, the distance between the two point clouds and corresponding points with ideal registrations are estimated theoretically, according to the range error and visual lateral resolution of the system. The simulation results show that the theoretically estimated results and actual results have the same variation tendency, thus providing a theoretical basis for subsequent improvements. Next, the estimated results are added according to the iterative closest point algorithm. This implies that the registration thresholds are capable of changing and adapting under different iterations and imaging systems, thus improving the speed and accuracy of registrations. This phenomenon is not seen in other algorithms. Experiments involving laser array imaging of a point cloud and laser scanning of depth imaging data show that the algorithm is practical and effective for both imaging types of point clouds and can improve the speed and accuracy of registration notably. The effectiveness and feasibility of the proposed algorithm are thus verified. In addition, for its full consideration of the imaging system, the basic idea of the proposed algorithm can be used for designing future applications as required.

There are many sources of point cloud data, such as the point cloud model obtained after a bundle adjustment of aerial images, the point cloud acquired by scanning a vehicle-borne light detection and ranging (LiDAR), the point cloud acquired by terrestrial laser scanning, etc. Different sensors use different processing methods. They have their own advantages and disadvantages in terms of accuracy, range and point cloud magnitude. Point cloud fusion can combine the advantages of each point cloud to generate a point cloud with higher accuracy. Following the classic Iterative Closest Point (ICP), a virtual namesake point multi-source point cloud data fusion based on Fast Point Feature Histograms (FPFH) feature difference is proposed. For the multi-source point cloud with noise, different sampling resolution and local distortion, it can acquire better registration effect and improve the accuracy of low precision point cloud. To find the corresponding point pairs in the ICP algorithm, we use the FPFH feature difference, which can combine surrounding neighborhood information and have strong robustness to noise, to generate virtual points with the same name to obtain corresponding point pairs for registration. Specifically, through the establishment of voxels, according to the F2 distance of the FPFH of the target point cloud and the source point cloud, the convolutional neural network is used to output a virtual and more realistic and theoretical corresponding point to achieve multi-source Point cloud data registration. Compared with the ICP algorithm for finding corresponding points in existing points, this method is more reasonable and more accurate, and can accurately correct low-precision point cloud in detail. The experimental results show that the accuracy of our method and the best algorithm is equivalent under the clean point cloud and point cloud of different resolutions. In the case of noise and distortion in the point cloud, our method is better than other algorithms. For low-precision point cloud, it can better match the target point cloud in detail, with better stability and robustness.

Aiming at the insufficient accuracy and the low efficiency of 3D point cloud registration, a point cloud registration method based on subgraph matching and reinforcement learning was proposed to achieve the accurate and fast registration of low-quality point cloud. Firstly, the 3D point cloud registration can result from a series of discrete rigid transformation actions, and this work used a reinforcement learning strategy to train an end-to-end model to iteratively predict the rigid transformation actions. Then, for the model architecture, a Siamese backbone was used to extract the local feature information of the source point cloud and the target point cloud, respectively. Similar nodes in the source graph and the target graph were associated through a proposed cross-graph attention module. The aggregation of graph nodes was designed to extract global features of two graphs, by using the weighted sum with gating vectors. Finally, the global features of the source graph and the target graph were fused, and the discrete rigid transformation action was predicted based on the fused feature. The reinforcement learning strategy significantly improves the generalization of point cloud registration. The cross-graph attention module further improves the accuracy and efficiency of point cloud registration. Extensive experiments on both synthetic and real-scanned datasets, ModelNet40 and ScanObjectNN, demonstrate that, compared with the latest point cloud registration method, ReAgent, the proposed method can reduce the mean average error of rotation by at least 0.16 and the isotropic rotation error by at least 0.16, effectively improving the accuracy of registration on low-quality point clouds.

Point cloud registration is a fundamental problem in 3D computer vision. In this paper, we cast point cloud registration into a planning problem in reinforcement learning, which can seek the transformation between the source and target point clouds through trial and error. By modeling the point cloud registration process as a Markov decision process (MDP), we develop a latent dynamic model of point clouds, consisting of a transformation network and evaluation network. The transformation network aims to predict the new transformed feature of the point cloud after performing a rigid transformation (i.e., action) on it while the evaluation network aims to predict the alignment precision between the transformed source point cloud and target point cloud as the reward signal. Once the dynamic model of the point cloud is trained, we employ the cross-entropy method (CEM) to iteratively update the planning policy by maximizing the rewards in the point cloud registration process. Thus, the optimal policy, i.e., the transformation between the source and target point clouds, can be obtained via gradually narrowing the search space of the transformation. Experimental results on ModelNet40 and 7Scene benchmark datasets demonstrate that our method can yield good registration performance in an unsupervised manner.

It has become an important and difficult point to solve the problem of non-cooperative target pose measurement in space. The three-dimensional point cloud data of the target acquired by the laser radar can restore its three-dimensional morphology through point cloud registration, so as to solve the relative pose of non-cooperative target in space. In view of the fact that the traditional point cloud Iterative Closest Point (ICP) registration algorithm is prone to fall into local extremum when the initial value is not good, which leads to registration failure, a rough registration algorithm of 3D point cloud based on the combination of bounding box and FPFH descriptor using principal component analysis is proposed. In this method, PCA is used to construct a three-dimensional point cloud bounding box. The parameter threshold of FPFH describing sub-point cloud on-time matching is selected through the density of point cloud and the outer dimension of outer bounding box. The normal alignment is added to adjust the normal direction of point cloud. The experimental results show that the proposed method simplifies the complexity feature parameters selection, reduces the search range of the feature point matching, and provides the error of precise registration of about 3°, to solve the spatial non-cooperative target relative position measurement technology provides technical reference.

Self-supervised rigid transformation equivariance for accurate 3D point cloud registration