Iterative Closest Point (ICP)

In this paper, we present an automated system for the recognition and identification of fingerprints based on rigid registration algorithms. Indeed, after preprocessing carried on a fingerprint database collected in the laboratory, we have built maps of minutiae for each fingerprint. Subsequently, we applied a rigid registration algorithm based on iterative search for closed points ICP (Iterative Closest Point), which allowed us to compare shifted fingerprints serving as test with the fingerprints of the reference database. This comparison gives convincing results and shows high accuracy.

The inspection of patients’ soft tissues and the effects of various dental procedures on their facial physiognomy are quite challenging. To minimise discomfort and simplify the process of manual measuring, we performed facial scanning and computer measurement of experimentally determined demarcation lines. Images were acquired using a low-cost 3D scanner. Two consecutive scans were obtained from 39 participants, to test the scanner repeatability. An additional ten persons were scanned before and after forward movement of the mandible (predicted treatment outcome). Sensor technology that combines red, green, and blue (RGB) data with depth information (RGBD) integration was used for merging frames into a 3D object. For proper comparison, the resulting images were registered together, which was performed with ICP (Iterative Closest Point)-based techniques. Measurements on 3D images were performed using the exact distance algorithm. One operator measured the same demarcation lines direc...

In 2010, he graduated with a PhD thesis on reliable carrier phase positioning (summa cum laude) and is now working towards his habilitation in the field of precise point positioning. He visited the Mathematical Geodesy and Positioning group at TU Delft in 2007, and the GPS Lab at Stanford University in 2008 and 2010. Dr. Henkel received the Pierre Contensou Gold Medal in 2007, the 1st prize in Bavaria at the European Satellite Navigation Competition in 2010, and the Vodafone Award for his dissertation in 2011. He is one of the founders and currently also the managing director of Advanced Navigation Solutions -ANavS.

Deep Learning-based 2D/3D registration methods are highly robust but often lack the necessary registration accuracy for clinical application. A refinement step using the classical optimization-based 2D/3D registration method applied in combination with Deep Learningbased techniques can provide the required accuracy. However, it also increases the runtime. In this work, we propose a novel Deep Learning driven 2D/3D registration framework that can be used end-to-end for iterative registration tasks without relying on any further refinement step. We accomplish this by learning the update step of the 2D/3D registration framework using Point-to-Plane Correspondences. The update step is learned using iterative residual refinement-based optical flow estimation, in combination with the Point-to-Plane correspondence solver embedded as a known operator. Our proposed method achieves an average runtime of around 8s, a mean re-projection distance error of 0.60 ± 0.40 mm with a success ratio of 97 percent and a capture range of 60 mm. The combination of high registration accuracy, high robustness, and fast runtime makes our solution ideal for clinical applications.

A method for segmenting three-dimensional scans of underwater unstructured terrains is presented. Individual terrain scans are represented as an elevation map and analysed using fast Fourier transform (FFT). The segmentation of the ground surface is performed in the frequency domain. The lower frequency components represent the slower varying undulations of the underlying ground whose segmentation is similar to de-noising / low pass filtering. The cut-off frequency, below which ground frequency components are selected, is automatically determined using peak detection. The user can specify a maximum admissible size of objects (relative to the extent of the scan) to drive the automatic detection of the cut-off frequency. The points above the estimated ground surface are clustered via standard proximity clustering to form object segments. The approach is evaluated using ground truth hand labelled data. It is also evaluated for registration error when the segments are fed as features to an alignment algorithm. In both sets of experiments, the approach is compared to three other segmentation techniques. The results show that the approach is applicable to a range of different terrains and is able to generate features useful for navigation. N x af ter -N x bef ore N x af ter

In this paper, we propose an approach to construct highly accurate 3D object models from range data. The main advantage of sensor based model acquisition compared to manual CAD model construction is the short time needed per object. The usual drawbacks of sensor based model reconstruction are sensor noise and errors in the sensor positions which typically lead to less accurate models. Our method drastically reduces this problem by applying a physical model of the underlying range sensor and utilizing a graph-based optimization technique. We present our approach and evaluate it on data recorded in different real world environments with an RGBD camera and a laser range scanner. The experimental results demonstrate that our method provides more accurate maps than standard SLAM methods and that it additionally compares favorable over the moving least squares method.

Image-guided surgery provides navigational assistance to the surgeon by displaying the surgical probe position on a set of preoperative tomograms in real time. In this study, the feasibility of implementing image-guided surgery concepts into liver surgery was examined during eight hepatic resection procedures. Preoperative tomographic image data were acquired and processed.

A simple extension method of the ICP algorithm using point quality based on the distance and incident angle is presented for improving the registration accuracy of TLS point clouds. First, using a structured point cloud representation, the distance and incident angle of each point are calculated. Then, quality measures for the distance and incident angle are defined, considering the relationship between them and the scanning errors. Finally, an extension method of the ICP algorithm by the weighting using the estimated point qualities is described. A method for efficient matching using point properties is also introduced. The proposed method is applied to two datasets of TLS point clouds of a small plant and an outdoor environment, and the effectiveness is evaluated by comparing registration errors with the ground truths obtained using the scan target and a total station.

A simple extension method of the ICP algorithm using point quality based on the distance and incident angle is presented for improving the registration accuracy of TLS point clouds. First, using a structured point cloud representation, the distance and incident angle of each point are calculated. Then, quality measures for the distance and incident angle are defined, considering the relationship between them and the scanning errors. Finally, an extension method of the ICP algorithm by the weighting using the estimated point qualities is described. A method for efficient matching using point properties is also introduced. The proposed method is applied to two datasets of TLS point clouds of a small plant and an outdoor environment, and the effectiveness is evaluated by comparing registration errors with the ground truths obtained using the scan target and a total station.

Recently, by Mobile Mapping System (MMS) with laser scanners, a GPS and IMU (Inertial Measurement Unit), 3D point clouds of urban areas (MMS point clouds) are easily acquired. When the same areas are scanned several times by the MMS, the point clouds often have differences in the range of several hundreds of millimetres. Such differences are caused by inertial drifts of IMU and losses of GPS signals in urban areas. In this paper, we propose an automatic accurate registration method of MMS point clouds using a new variant of ICP (Iterative Closest Point) algorithm for MMS point clouds and trajectory modification. Our method consists of four steps. Firstly, some trajectory points are automatically extracted by analyzing the trajectory. Secondly, the differences of point clouds are derived at the extracted trajectory points in the overlapping scan region by our new ICP algorithm which minimizes pointto-plane and point-to-point distances simultaneously and filters incorrect correspondences based on a point classification by PCA (Principle Component Analysis). Thirdly, the modified positions and rotation parameters at all extracted trajectory points are derived by a least squares method for positioning and registration constraints. Finally, each point in the point clouds is modified by coordinate transformations which are derived from linear interpolation of the modified positions and rotation parameters of the extracted trajectory points. Our method was applied to MMS point clouds and trajectory and the performances were evaluated.

Service robots can benefit from maps that support their tasks and facilitate communication with humans. For efficient interaction, it is practical to be able to reference structures and objects in the environment, e.g. "fetch the mug from the kitchen table." Towards this end, we present a feature-based SLAM and semantic mapping system which uses a variety of feature types as landmarks, including planar surfaces such as walls, tables, and shelves, as well as objects such as door signs. These landmarks can be optionally labeled by a human for later reference. Support for partitioning maps into labeled regions or places is also presented.

Many previous literatures use landmarks to guide the correspondence of 3D faces. However, these landmarks, either manually or automatically annotated, are hard to define consistently across different faces in many circumstances. We propose a general framework for dense correspondence of 3D faces without landmarks in this paper. The dense correspondence goal is revisited in two perspectives: semantic and topological correspondence. Starting from a template facial mesh, we sequentially perform global alignment, primary correspondence by template warping, and contextual mesh refinement, to reach the final correspondence result. The semantic correspondence is achieved by a local iterative closest point (ICP) algorithm of kernelized version, allowing accurate matching of local features. Then, robust deformation from the template to the target face is formulated as a minimization problem. Furthermore, this problem leads to a well-posed sparse linear system such that the solution is unique and efficient. Finally, a contextual mesh refining algorithm is applied to ensure topological correspondence. In the experiment, the proposed method is evaluated both qualitatively and quantitatively on two datasets including a publicly available FRGC v2.0 dataset, demonstrating reasonable and reliable correspondence results.

In this technical report we present a data set which can be used by researchers to test and compare their algorithms related to feature-based SLAM and in particular data association tasks. The data set covers a large path mostly through an indoor office environment with artificial and natural landmarks. It provides odometry and geometrical results from computer vision-based landmark detection, as well as raw images and camera calibration data. The artificial landmarks are white discs on the floor where the position relative to the robot is provided as measurement. The natural landmarks are vertical lines where only the bearing relative to the robot is provided and data association is remarkably difficult. This technical report provides full information about experimental setup, model and measurement functions. The focus of this data set is on data association and for comparison of different approaches we provide full ground truth by annotating all features.

This paper proposes a camera projection model for an oblique‐viewing endoscope, intended for application to an extended virtual reality system. The oblique‐viewing endoscope differs from the forward‐viewing endoscope by the fact that the view field can be shifted by rotating the scope cylinder. The projection model proposed in this study describes the systematic relation between the rotation angle of the scope cylinder and the external parameters of the camera model. A calibration method is also proposed for estimating the parameters of the projection model, and its accuracy is investigated. The oblique‐viewing projection model is derived as an extension of the general pinhole camera model by adding two axes of rotation as parameters and using the angle of rotation of the scope cylinder as the input. In the experiment, an abdominal endoscope is used as the real equipment. A rotary encoder with an angular resolution of 0.25° is used to input the rotation angle of the scope cylinder. ...

In this paper we present an algorithm for groupwise image alignment using an iterative best edge point algorithm. Neighbouring image edge points are matched for similarity in a two directional fashion. The matches found are used to drive a regularised warp of the images into alignment. The algorithm works from low to high resolution, with the matches calculated across the set first at low resolution and towards progressively finer scales. The regularisation decreases across iterations, and the search area remains constant, so covers larger effective area in the low resolution images. We also extend the method to 3D surfaces by combining the 2D image search with a 3D ICP algorithm. The results show that this gives a very efficient algorithm that can align many different sets of 2D images and 3D surfaces. 22 Tiddeman B. and Hunter D.. GROUPWISE NON-RIGID IMAGE ALIGNMENT -Results of a Multiscale Iterative Best Edge Algorithm.

Digital stereo-photogrammetry allows users an automatic evaluation of the spatial dimension and the surface texture of objects. The integration of image analysis techniques simplifies the automation of evaluation of large image sets and offers a high accuracy . Due to the substantial similarities of stereoscopic image pairs, correlation techniques provide measurements of subpixel precision for corresponding image points. With the help of an automated point search algorithm in image sets identical points are used to associate pairs of images to stereo models and group them. The found identical points in all images are basis for calculation of the relative orientation of each stereo model as well as defining the relation of neighboured stereo models. By using proper filter strategies incorrect points are removed and the relative orientation of the stereo model can be made automatically. With the help of 3D-reference points or distances at the object or a defined distance of camera basis the stereo model is orientated absolute. An adapted expansion-and matching algorithm offers the possibility to scan the object surface automatically. The result is a three dimensional point cloud, the scan resolution depends on image quality. With the integration of the iterative closest point-algorithm (ICP) these partial point clouds are fitted to a total point cloud. In this way, 3D-reference points are not necessary. With the help of the implemented triangulation algorithm a digital surface models (DSM) can be created. The texturing can be made automatically by the usage of the images, that were used for scanning the object surface. It is possible to texture the surface model directly or to generate orthophotos automatical. By using of calibrated digital SLR cameras with full frame sensor a high accuracy can be reached. A big advantage is the possibility to control the accuracy and quality of the 3d-objectdocumentation with the resolution of the images. The procedure described here is implemented in software metigo 3D.

Digital stereo-photogrammetry allows users an automatic evaluation of the spatial dimension and the surface texture of objects. The integration of image analysis techniques simplifies the automation of evaluation of large image sets and offers a high accuracy [1]. Due to the substantial similarities of stereoscopic image pairs, correlation techniques provide measurements of subpixel precision for corresponding image points. With the help of an automated point search algorithm in image sets identical points are used to associate pairs of images to stereo models and group them. The found identical points in all images are basis for calculation of the relative orientation of each stereo model as well as defining the relation of neighboured stereo models. By using proper filter strategies incorrect points are removed and the relative orientation of the stereo model can be made automatically. With the help of 3D-reference points or distances at the object or a defined distance of camera ...

Digital Smile Design (DSD) and Computer-Aided Design/Computer-Aided Manufacturing (CAD/CAM) technologies have transformed aesthetic and restorative dentistry, offering unprecedented precision in diagnostics, treatment planning, and patient communication. However, their benefits remain disproportionately concentrated in affluent urban areas. This paper examines how the strategic implementation of costeffective digital tools-such as Medit and iTero intraoral scanners, AI-powered diagnostics, and teledentistry-can expand access to high-quality dental care for underserved communities across the United States. Evidence suggests these technologies can reduce treatment costs, shorten procedure times, and improve oral health equity when integrated into community-based clinics. The findings underscore the role of digital dentistry not only as a clinical innovation but also as a public health tool capable of addressing systemic disparities in oral healthcare delivery.

To improve the searching efficiency of the precise position of workpiece with free-form surface, a new two- step point cloud registration method is proposed. It seeks to improve the preliminary registration algorithm and the iterative closest point (ICP) algorithm. Firstly, for preliminary registration, the co-planar 4-points sets are searched based on the constraints of distance and proportional relationship between points. Then the invariant of these co-planar 4-points sets are used for random sample consensus algorithm (RANSAC) to get initial position of the target point cloud. Secondly, re-sampling of source data, matching method based on double direction normal projection and elimination of unreliable point-pairs are used to improve the original ICP algorithm to optimize the preliminary registration result. The registration simulations of two simple workpieces were conducted. The results show that the improved algorithm has higher efficiency and better precision than the tradit...

We describe a variation of the iterative closest point (ICP) algorithm for aligning two point sets under a set of transformations. Our algorithm is superior to previous algorithms because (1) in determining the optimal alignment, it identifies and discards likely outliers in a statistically robust manner, and (2) it is guaranteed to converge to a locally optimal solution. To this end, we formalize a new distance measure, fractional root mean squared distance (FRMSD), which incorporates the fraction of inliers into the distance function. We lay out a specific implementation, but our framework can easily incorporate most techniques and heuristics from modern registration algorithms. We experimentally validate our algorithm against previous techniques on 2 and 3 dimensional data exposed to a variety of outlier types.

The utilization of terrestrial laser scanning (TLS) and airborne laser scanning (ALS ) point cloud data in forest inventory studies has significantly increased. Fusing of TLS and ALS point cloud data has been recognized as an effective approach in forest studies. In this regard, co-registration of point clouds is considered one of the crucial steps in the integration process. Co-registering point clouds in forest environments faces various challenges, including unstable features, extensive occlusions, different viewpoints, and differences in point cloud densities. To address these intricate challenges, this study introduces an automated and robust method for co-registering TLS and ALS point clouds based on the correspondence of individual tree locations in forest environments. Recently, tree location-based methodologies have been advanced to grapple with these complexities in forest environments. However, many of these methods are highly sensitive to the accuracy of tree positioning. The proposed approach aims to reduce sensitivity to individual tree positioning accuracy. Initially, the positions of individual trees in both TLS and ALS data are extracted. Then, a filtering approach is applied to eliminate positions with low potential for corresponding matches in the TLS and ALS dataset. S ince larger trees in the TLS data have a higher potential for corresponding matches in the ALS data, an iterative process is applied to identify correspondences between trees in both datasets. After estimating transformation parameters, the coregistration process is executed. The proposed method is applied on six datasets with varying forest complexities. The results demonstrate a high success rate up to 100% if the starting position of the TLS plots are located within ~4 hectares (~2000 trees). Additionally, the potential of the proposed method for co-registering TLS data with ALS data across different search areas and varying number of trees is evaluated in detail. The outcomes indicate that successful co-registration of TLS plot with 50 m diameter to ALS data is successful in the best case within a search radius of approximately 113 hectares (~60,000 tree locations) and in the worst case for around 20 hectares (~10,000 tree locations) depending on the forest complexity. T his paragraph of the first footnote will contain the date on which you submitted your paper for review, which is populated by IEEE. It is IEEE style to display support information, including sponsor and financial support acknowledgment, here and not in an acknowledgment section at the end of the article. For example, "This work was supported in part by the U.S. Department of Commerce under Grant BS123456." The name of the corresponding author appears after the financial information, e.g.

In this paper, we intend to propose a framework for the description and the matching of three dimensional faces. Our starting point is the representation of the 3D face by an invariant description under the M (3) group of translations and rotations. This representation is materialized by the points of the arc-length reparametrization of all the level curves of the three polar representation. These points are indexed by their level curve number and their position in each level. With this type of description we need a step of registration to align 3D faces with different expressions. Therefore, we propose to use a robust version of the iterative closest point algorithm (ICP) adopted to 3D face recognition context. We test the accuracy of our approach on a part of the BU-3DFE database of 3D faces. The obtained results for many protocols of the identification scenario show the performance of such framework.

In this paper, we intend to propose a framework for the description and the matching of three dimensional faces. Our starting point is the representation of the 3D face by an invariant description under the M (3) group of translations and rotations. This representation is materialized by the points of the arc-length reparametrization of all the level curves of the three polar representation. These points are indexed by their level curve number and their position in each level. With this type of description we need a step of registration to align 3D faces with different expressions. Therefore, we propose to use a robust version of the iterative closest point algorithm (ICP) adopted to 3D face recognition context. We test the accuracy of our approach on a part of the BU-3DFE database of 3D faces. The obtained results for many protocols of the identification scenario show the performance of such framework.

In this research an innovative fully automated 3D face compression and recognition system is presented. Several novelties are introduced to make the system performance robust and efficient. These novelties include: First, an automatic pose correction and normalization process by using curvature analysis for nose tip detection and iterative closest point (ICP) image registration. Second, the use of spherical based wavelet coefficients for efficient representation of the 3D face. The spherical wavelet transformation is used to decompose the face image into multi-resolution sub images characterizing the underlying functions in a local fashion in both spacial and frequency domains. Two representation features based on spherical wavelet parameterization of the face image were proposed for the 3D face compression and recognition. Principle component analysis (PCA) is used to project to a low resolution sub-band. To evaluate the performance of the proposed approach, experiments were performed on the GAVAB face database. Experimental results show that the spherical wavelet coefficients yield excellent compression capabilities with minimal set of features. Haar wavelet coefficients extracted from the face geometry image was found to generate good recognition results that outperform other methods working on the GAVAB database.

Laser scan matching with grid-based maps is a promising tool for real-time indoor positioning of mobile Unmanned Ground Vehicles (UGVs). While there are critical implementation problems, such as the ability to estimate the position by sensing the unknown indoor environment with sufficient accuracy and low enough latency for stable vehicle control, further development work is necessary. Unfortunately, most of the existing methods employ heuristics for quick positioning in which numerous accumulated errors easily lead to loss of positioning accuracy. This severely restricts its applications in large areas and over lengthy periods of time. This paper introduces an efficient real-time mobile UGV indoor positioning system for large-area applications using laser scan matching with an improved probabilistically-motivated Maximum Likelihood Estimation (IMLE) algorithm, which is based on a multi-resolution patch-divided grid likelihood map. Compared with traditional methods, the improvements embodied in IMLE include: (a) Iterative Closed Point (ICP) preprocessing, which adaptively decreases the search scope; (b) a totally brute search matching method on multi-resolution map layers, based on the likelihood value between current laser scan and the grid map within refined search scope, adopted to obtain the global optimum position at each scan matching; and (c) a

This paper concerns implementation of algorithms in the two important aspects of modern 3D data processing: data registration and segmentation. Solution proposed for the first topic is based on the 3D space decomposition, while the latter on image processing and local neighbourhood search. Data processing is implemented by using NVIDIA compute unified device architecture (NIVIDIA CUDA) parallel computation. The result of the segmentation is a coloured map where different colours correspond to different objects, such as walls, floor and stairs. The research is related to the problem of collecting 3D data with a RGB-D camera mounted on a rotated head, to be used in mobile robot applications. Performance of the data registration algorithm is aimed for on-line processing. The iterative closest point (ICP) approach is chosen as a registration method. Computations are based on the parallel fast nearest neighbour search. This procedure decomposes 3D space into cubic buckets and, therefore, the time of the matching is deterministic. First technique of the data segmentation uses accelerometers integrated with a RGB-D sensor to obtain rotation compensation and image processing method for defining prerequisites of the known categories. The second technique uses the adapted nearest neighbour search procedure for obtaining normal vectors for each range point.

To assess if a given manufactured artifact conform to a template CAD model, the artifact is measured by a coordinate measuring system to obtain a point cloud. The point cloud is compared to the CAD model by fitting the point cloud to the CAD model. One of approaches is based on a method required to determine footpoint parameters. The footpoints are the projection of the points in the point cloud onto the CAD surface. In this paper we introduce a new efficient method for estimating footpoint parameters.

We propose the use of a "pick-up" ultrasound transducer for robot-assisted minimally invasive surgeries. Unlike prior approaches, the ultrasound transducer is inserted before the procedure and remains in the abdominal cavity throughout. We present a new design for such an intra-abdominal ultrasound transducer with a handle that can be grasped in a repeatable manner using a da Vinci Pro-Grasp tool. The main application is mapping the vasculature, which is segmented from Doppler and B-mode images using a Kalman-filtering approach. Our goal is employ the vasculature to register pre-operative CT to intra-operative camera images. To demonstrate the feasibility of the approach, we use an ultrasound flow phantom to register a CT surface model to extracted ultrasound vessel center points using an iterative closest point method. The transducer was tracked with electromagnetic sensors and a target registration error of 3.2 mm was calculated. The initial application will be nephrectomy where vessel localization is paramount.

Learning a map of an unknown environment and localising a robot in it is a common problem in robotics, with solutions usually requiring an estimate of the robot's motion. In scenarios such as Urban Search and Rescue, motion encoders can be highly inaccurate, and weight and battery requirements often limit computing power. We have developed a GPU based algorithm using Iterative Closest Point position tracking and Graph SLAM that can accurately generate a map of an unknown environment without the need for motion encoders and requiring minimal computational resources. The algorithm is able to correct for drift in the position tracking by rapidly identifying loops and optimising the map. We present a method for refining the existing map when revisiting areas to increase the accuracy of the existing map and bound the run-time to the size of the environment.

A n o vel model-based pose estimation algorithm is presented which estimates the motion of a three-dimensional object from an image sequence. The nonlinear estimation process within iteration is divided into two linear estimation stages, namely the depth approximation and the pose calculation. In the depth approximation stage, the depths of the feature points in three-dimensional space are estimated. In the pose calculation stage, the rotation and translation parameters between the estimated feature points and the model point set are calculated by a fast singular value decomposition method. The whole process is executed recursively until the result is stable. Since both stages can be solved e ciently, the computational cost is low. As a result, the algorithm is well-suited for real time computer vision applications. We demonstrate the capability o f t h i s algorithm by applying it to a real time head tracking problem. The results are satisfactory. keyword : Pose Estimation, singular value decomposition, real time vision and Human-computer Interface.

Image and point cloud registration methods compute the relative pose between two images. Commonly used registration algorithms are iterative and rely on the assumption that the motion between the images is small. In this work, we propose a fast pose estimation technique to compute a rough estimate of large motions between depth images, which can be used as initialization to dense registration methods. The main idea is to explore the properties given by planar surfaces with co-visibility and their normals from two distinct viewpoints. We present, in two decoupled stages, the rotation and then the translation estimation, both based on the normal vectors orientation and on the depth. These two stages are efficiently computed by using low resolution depth images and without any feature extraction/matching. We also analyze the limitations and observabilty of this approach, and its relationship to ICP point-to-plane. Notably, if the rotation is observable, at least five degrees of freedom can be estimated in the worst case. To demonstrate the effectiveness of the method, we evaluate the initialization technique in a set of challenging scenarios, comprising simulated spherical images from the Sponza Atrium model benchmark and real spherical indoor sequences.

Image and point cloud registration methods compute the relative pose between two images. Commonly used registration algorithms are iterative and rely on the assumption that the motion between the images is small. In this work, we propose a fast pose estimation technique to compute a rough estimate of large motions between depth images, which can be used as initialization to dense registration methods. The main idea is to explore the properties given by planar surfaces with co-visibility and their normals from two distinct viewpoints. We present, in two decoupled stages, the rotation and then the translation estimation, both based on the normal vectors orientation and on the depth. These two stages are efficiently computed by using low resolution depth images and without any feature extraction/matching. We also analyze the limitations and observabilty of this approach, and its relationship to ICP point-to-plane. Notably, if the rotation is observable, at least five degrees of freedom can be estimated in the worst case. To demonstrate the effectiveness of the method, we evaluate the initialization technique in a set of challenging scenarios, comprising simulated spherical images from the Sponza Atrium model benchmark and real spherical indoor sequences.

This paper presents a method for measuring the in-bucket payload volume on a dragline excavator for the purpose of estimating the material's bulk density in realtime. Knowledge of the payload's bulk density can provide feedback to mine planning and scheduling to improve blasting and therefore provide a more uniform bulk density across the excavation site. This allows a single optimal bucket size to be used for maximum overburden removal per dig and in turn reduce costs and emissions in dragline operation and maintenance. The proposed solution uses a range bearing laser to locate and scan full buckets between the lift and dump stages of the dragline cycle. The bucket is segmented from the scene using cluster analysis, and the pose of the bucket is calculated using the Iterative Closest Point (ICP) algorithm. Payload points are identified using a known model and subsequently converted into a height grid for volume estimation. Results from both scaled and full scale implementations show that this method can achieve an accuracy of above 95%.

This paper presents a method for measuring the in-bucket payload volume on a dragline excavator for the purpose of estimating the material's bulk density in realtime. Knowledge of the payload's bulk density can provide feedback to mine planning and scheduling to improve blasting and therefore provide a more uniform bulk density across the excavation site. This allows a single optimal bucket size to be used for maximum overburden removal per dig and in turn reduce costs and emissions in dragline operation and maintenance. The proposed solution uses a range bearing laser to locate and scan full buckets between the lift and dump stages of the dragline cycle. The bucket is segmented from the scene using cluster analysis, and the pose of the bucket is calculated using the Iterative Closest Point (ICP) algorithm. Payload points are identified using a known model and subsequently converted into a height grid for volume estimation. Results from both scaled and full scale implementations show that this method can achieve an accuracy of above 95%.

We present an approach for estimating the relative transformations between fragments of a broken cylindrical structure in 3d. To solve this problem, we first measure the orientation and position of the cylinder axes for each fragment by an adapted kind of Hough Transformation. The cylinder axes are an important feature for separation of fractured areas and for calculation of an initial reposition solution (constraining 4 DOFs). After these processing steps, we compute the relative transformations between corresponding fragments by using well-known surface registration techniques, like 2d depths correlation and the ICP (Iterative Closest Point) algorithm. One goal of our project is to use the proposed method for estimation of relative transformations between fragments of fractured long bones for computer aided and semi-automatic bone alignment and fracture reduction in surgery.

In this paper we present a new variant of ICP (iterative closest point) algorithm based on local feature correlation. Our approach combines global and local feature information to find better correspondence sets and to use them to compute the 3D pose of the object model even for the case of large displacements between model and image data. For such cases, we propose a 2D alignment in the image plane (rotation plus translation) before the feature extraction process. This has some advantages over the classical methods like better convergence and robustness. Furthermore, it avoids the need of a normal pre-alignment step in 3D. Our approach was tested on synthetical and real-world data to compare the convergence behavior and performance against other versions of the ICP algorithm combined with a classical pre-alignment approach.

The utilization of terrestrial laser scanning (TLS) and airborne laser scanning (ALS) point cloud data in forest inventory studies has significantly increased. Fusing of TLS and ALS point cloud data has been recognized as an effective approach in forest studies. In this regard, co-registration of point clouds is considered one of the crucial steps in the integration process. Co-registering point clouds in forest environments faces various challenges, including unstable features, extensive occlusions, different viewpoints, and differences in point cloud densities. To address these intricate challenges, this study introduces an automated and robust method for co-registering TLS and ALS point clouds based on the correspondence of individual tree locations in forest environments. Recently, tree location-based methodologies have been advanced to grapple with these complexities in forest environments. However, many of these methods are highly sensitive to the accuracy of tree positioning. The proposed approach aims to reduce sensitivity to individual tree positioning accuracy. Initially, the positions of individual trees in both TLS and ALS data are extracted. Then, a filtering approach is applied to eliminate positions with low potential for corresponding matches in the TLS and ALS dataset. S ince larger trees in the TLS data have a higher potential for corresponding matches in the ALS data, an iterative process is applied to identify correspondences between trees in both datasets. After estimating transformation parameters, the coregistration process is executed. The proposed method is applied on six datasets with varying forest complexities. The results demonstrate a high success rate up to 100% if the starting position of the TLS plots are located within ~4 hectares (~2000 trees). Additionally, the potential of the proposed method for co-registering TLS data with ALS data across different search areas and varying number of trees is evaluated in detail. The outcomes indicate that successful co-registration of TLS plot with 50 m diameter to ALS data is successful in the best case within a search radius of approximately 113 hectares (~60,000 tree locations) and in the worst case for around 20 hectares (~10,000 tree locations) depending on the forest complexity.

Figure 2. (a) and (b) show two middle cerebral artery trees are extracted and overlapped on the MIPs. Vessel trees are visualized in (c) and (d) with a constant radius of 0.2mm. This infant was scanned at the age of 7 days and 63 days. (c) Before registration, significant misalignment can be observed for two composed vessel trees. (d) Indirect registration method effectively decreases the misalignment between two vessel trees, which is necessary for correct branch matching. Local deformation left can be further reduced after the subtree registration. (a) (c) (d)

In this paper, we investigate the impact of different kind of car trajectories on LiDAR scans. In fact, LiDAR scanning speeds are considerably slower than car speeds introducing distortions. We propose a method to overcome this issue as well as new metrics based on CAN bus data. Our results suggest that the vehicle trajectory should be taken into account when building 3D large-scale maps from a LiDAR mounted on a moving vehicle.

In this paper, we investigate the impact of different kind of car trajectories on LiDAR scans. In fact, LiDAR scanning speeds are considerably slower than car speeds introducing distortions. We propose a method to overcome this issue as well as new metrics based on CAN bus data. Our results suggest that the vehicle trajectory should be taken into account when building 3D large-scale maps from a LiDAR mounted on a moving vehicle.

Localization and mapping are very important for safe movement of robots. One possible way to assist with this functionality is to use laser scan matching. This paper describes a method to implement this functionality. It is based on well-known random sampling and consensus (RANSAC) and iterative closest point (ICP). The proposed algorithm belongs to the class of point to point scan matching approach with its matching criteria rule. The performance of the proposed algorithm is examined in real environment and found applicable in real-time application.

Shape registration is a key stage in the process of reconstruction or acquisition of three-dimensional surfaces so that it is considered a cornerstone in the fields of computer vision and graphics. It is also a task of major importance in many other fields like biometric applications [1], 3D model reconstruction from multiple range images [2], medical diagnostic support tools and computer aided interventions [3], quality control of manufactured pieces [4] and robotic applications ranging from full environment reconstruction for navigation to particular objects 3D model creation and tracking [5]. A remarkable study of the different registration techniques both rigid and non-rigid is the work by Tam et al. [6] which includes an in depth review of the registration problem; other important surveys are the ones by Van Kaik et al. [7] (focused on shape correspondence techniques) and Audette et al. [3] (centered on shape registration applied to medical images). In addition, registration ca...

The most commonly used method for addressing 3D geometric registration is the iterative closet-point algorithm, this approach is incremental and prone to drift over multiple consecutive frames. The Common strategy to address the drift is the pose graph optimization subsequent to frame-toframe registration, incorporating a loop closure process that identifies previously visited places. In this paper, we explore a framework that replaces traditional geometric registration and pose graph optimization with a learned model utilizing hierarchical attention mechanisms and graph neural networks. We propose a strategy to condense the data flow, preserving essential information required for the precise estimation of rigid poses. Our results, derived from tests on the KITTI Odometry dataset, demonstrate a significant improvement in pose estimation accuracy. This improvement is especially notable in determining rotational components when compared with results obtained through conventional multi-way registration via pose graph optimization. The code will be made available upon completion of the review process.

Bu bildiride, ifadenin varlığında sabit kayıtlamaya dayalı 3B yüz tanıma yöntemlerinin iyi başarım sağlamadığı gösterilmiş olup, başarım arttırımı için bölgesel kayıtlama ve tanıma yapılması önerilmiştir. Hızlı hizalama amaçlı, ortalama bölge modelleriyle yinelemeli en yakın nokta (ICP) algoritmasına dayalı kayıtlama yöntemi kullanılmıştır. Bölgelerden elde edilen benzerlik skorları, tümleştirme yöntemleriyle birleştirilerek başarım arttırımı sağlanmıştır. Bu çalışmada, ifade tanıma ve ifade varlığında yüz tanıma için kullanılmak üzere toplanmış yeni bir 3B yüz veri kütüphanesi tanıtılmış ve deneylerde bu veri kütüphanesi kullanılmıştır.

Localisation is essential for autonomous mobile robot system enabling it to locate itself within its environment. One method to perform localisation is to use scan matching with iteration closest point (ICP) algorithm. However, typical ICP may be prone to inaccuracies in localisation and mapping due to problems associated with laser range data limitation such as overshoot data and blank data. This paper presents the improvement to the above problem by the inclusion of a threshold to the KNN scan matching algorithm during iteration process. The threshold is a percentage of nearest point of incoming input with respected to reference point. Threshold values of 0%, 70% and 90% were tested, and improvements of the classification performance were observed with the increase in the threshold values, with the latter achieving 100% accuracy. This work shows that the use of threshold in scan matching may improve the accuracy of local map classification.

In this paper we present a novel approach to quickly obtain detailed 3D reconstructions of large scale environments. The method is based on the consecutive registration of 3D point clouds generated by modern lidar scanners such as the Velodyne HDL-32e or HDL-64e. The main contribution of this work is that the proposed system specifically deals with the problem of sparsity and inhomogeneity of the point clouds typically produced by these scanners. More specifically, we combine the simplicity of the traditional iterative closest point (ICP) algorithm with the analysis of the underlying surface of each point in a local neighbourhood. The algorithm was evaluated on our own collected dataset captured with accurate ground truth. The experiments demonstrate that the system is producing highly detailed 3D maps at the speed of 10 sensor frames per second.

Airborne and terrestrial laser scanning and close-range photogrammetry are frequently used for very high-resolution mapping of land surface. These techniques require a good strategy of mapping to provide full visibility of all areas otherwise the resulting data will contain areas with no data (data shadows). Especially, deglaciated rugged alpine terrain with abundant large boulders, vertical rock faces and polished roche-moutones surfaces complicated by poor accessibility for terrestrial mapping are still a challenge. In this paper, we present a novel methodological approach based on a combined use of terrestrial laser scanning (TLS) and close-range photogrammetry from an unmanned aerial vehicle (UAV) for generating a high-resolution point cloud and digital elevation model (DEM) of a complex alpine terrain. The approach is demonstrated using a small study area in the upper part of a deglaciated valley in the Tatry Mountains, Slovakia. The more accurate TLS point cloud was supplemented by the UAV point cloud in areas with insufficient TLS data coverage. The accuracy of the iterative closest point adjustment of the UAV and TLS point clouds was in the order of several centimeters but standard deviation of the mutual orientation of TLS scans was in the order of millimeters. The generated high-resolution DEM was compared to SRTM DEM, TanDEM-X and national DMR3 DEM products confirming an excellent applicability in a wide range of geomorphologic applications.

The analysis and estimation is an essential step towards to a reliable reconstruction process and for a proper use of 3D data. Therefore it is interesting to analyze which error sources influence the final result, and what is the sensitivity of each of these error sources. The objective of this work is to perform the error analysis of a reconstructed surface. The 3D reconstructed scene has the precision improved by combining point clouds obtained from different viewpoints using structured light. The main task is the point cloud registration algorithm that matches two point clouds. A well-known algorithm for point cloud registration is the ICP that determines the rotation and translation that when applied to one of the point clouds, place both point clouds in accordance. An improved ICP is used in the error analysis. The quality of the 3D reconstruction is described as a function of input data and the correctness of the reconstructed scene is compared with the real surface measurement.