Papers by Keyword: Point Cloud

Abstract: Most of types of laser distance measuring instrument cost hundreds of thousand dollars such as Atos scanner or Depth Camera that gives depth maps of space very fast. However, the handling is too complicated for non-professional users and the utilization of 3D reconstruction is very limited. This paper introduces a workflow of 3D reconstruction using a new cheaper instrument, the Microsoft Kinect. The first experiences with Microsoft Kinect v2 sensor are presented, and the ability of 3D modelling for mechanical parts is investigated. For this purpose, the point cloud on output data as well as a calibration approach are demonstrated.

Abstract: This paper presents a novel structure-based registration method for terrestrial laser scanning (TLS) data. The line support region (LSR), which fits the 3D line segment, is adopted to describe the scene structure and reduce geometric complexity. Then we employ an evolution computation method to solve the optimization problem of global registration. Our method can be further enhanced by iterative closest points (ICP) or other local registration methods. We demonstrate the robustness of our algorithm on several point cloud sets with varying extent of overlap and degree of noise.

Abstract: For some special blade section, an accurate partitioning algorithm for feature point cloud is presented to avoid the drawbacks of traditional convex hull algorithm. On this basis, the method based on graph transformation is used to extract the chord length. The experimental results show that these two algorithms have strong generality, high accuracy and efficiency, can meet the testing requirements of blades.

Abstract: Engineers and designers constantly strive to design machines that have high life cycle, can run faster, and can manufacture products having higher geometrical accuracy. However, manufacturing processes often produce parts with surfaces that are unsatisfactory from the viewpoint of geometrical precision or quality of surface texture. In general, electronic, optical, and tactual methods are used for exploring and evaluating the surface texture and the surface geometry. With the advancements in the reverse engineering technology, computer based numerical analysis of the data obtained from 3D imaging devices, referred to as registration, has also emerged as a tool for shape inspection. Registration expedites shape inspection by augmenting the process of error measurement. The existing registration methods suffer from problems that limit the extensive use of the existing approaches, thereby making them application specific. In this research, we propose a novel approach for reducing the number of data points used for registration without compromising with registration accuracy. We propose a segmentation based registration approach that utilizes the features extracted from the models, thus reducing the data size. The registered data sets can be compared to each other for evaluating the differences in shape. The difference information can further be used to appropriately modify the manufacturing process parameters for obtaining highly accurate products.

Abstract: For the problem of huge computation and requiring high computing resource in point cloud registration, according to the theory of parallel computing, the algorithm of point cloud registration base on MapReduce is designed. Through building a Hadoop cluster consisted by average PCs, four examples have been tested. The experiment results show that point cloud registration algorithm based on MapReduce can register point cloud data with high accuracy.

Abstract: For vehicle-borne LiDAR, a mathematical model is built for the computation and reconstruction of laser point cloud with the scanning data, GPS data and IMU data. 3D point cloud of the road and the scenery on the both sides of the road is obtained. Then according to the trajectory of the vehicle, 3D roaming for the scenery on the both sides of the road is realized using OpenGL 3D engine technology. This technology provides a probably feasible way for anti-collision of vehicles and aircrafts when driven at night, in the heavy fog or flying between the mountains.

Abstract: Point cloud has been used as representation model for tool path planning. However, for point cloud machining, there are few tool path generation strategies but iso-planar method. In this paper, a method of generating tool paths from point cloud for three-axis machining is presented. A simple base surface, such as plane or cylindrical surface, etc., is first created according to the shape of the point cloud. On this base surface, the guide curves of tool path are rapidly designed, and then are projected onto the point cloud surface to generate the final tool path by using the least squared projection method. Finally, this method is tested on a sample surface to demonstrate its validity.

Abstract: The objective of this work is to measure the residual stress normal to the cut plane which has a large size using contour method. Translations and rotations are applied to the second point cloud to assure the continuity of the contour. There is still the same work on the opposing cut plane, align and combine the two contour datasets, perform the interpolation and then integrate with a finite element modelling package to apply many thousands of different boundary conditions. The finite element result shows a good continuity indicating the success of transformation means applied on the point cloud. The result gets the stress distribution normal to the cut plane. Beam forming is simulated to study the residual stress distribution of the stretch-wrap bend forming, the residual stress from contour method and forming simulation present the same trend.

Abstract: Curvature estimation of 3-dimension discrete points performs an important role in dealing with scan line point cloud and is difficult to calculate. A discrete curvature estimation method based on local space parabola is proposed. Method in this paper is contrasted with circular arc fitting method and simulation experiment shows that the proposed method is feasible and effective with high precision.

Abstract: Some manufacturers of 3D digitizing systems are developing and market more accurate, fastest and affordable systems of fringe projection based on blue light technology. The aim of the present work is the determination of the quality and accuracy of the data provided by the LED structured light scanner Comet L3D (Steinbichler). The quality and accuracy of the cloud of points produced by the scanner is determined by measuring a number of gauge blocks of different sizes. The accuracy range of the scanner has been established through multiple digitizations showing the dependence on different factors such as the characteristics of the object and scanning procedure. Although many factors influence, accuracies announced by manufacturer have been achieved under optimal conditions and it has been noted that the quality of the point clouds (density, noise, dispersion of points) provided by this system is higher than that obtained with laser technology devices.