DOCSLIB.ORG

Real-Time Scene Reconstruction and Triangle Mesh Generation Using Multiple RGB-D Cameras Siim Meerits, Vincent Nozick, Hideo Saito

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Real-Time Scene Reconstruction and Triangle Mesh Generation Using Multiple RGB-D Cameras Siim Meerits, Vincent Nozick, Hideo Saito Real-time scene reconstruction and triangle mesh generation using multiple RGB-D cameras Siim Meerits, Vincent Nozick, Hideo Saito To cite this version: Siim Meerits, Vincent Nozick, Hideo Saito. Real-time scene reconstruction and triangle mesh genera- tion using multiple RGB-D cameras. Journal of Real-Time Image Processing, Springer Verlag, 2019, 10.1007/s11554-017-0736-x. hal-01638241 HAL Id: hal-01638241 https://hal-upec-upem.archives-ouvertes.fr/hal-01638241 Submitted on 26 Dec 2017 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Real-time scene reconstruction and triangle mesh generation using multiple RGB-D cameras Siim Meerits, Vincent Nozick & Hideo Saito Journal of Real-Time Image Processing ISSN 1861-8200 J Real-Time Image Proc DOI 10.1007/s11554-017-0736-x 1 23 Your article is published under the Creative Commons Attribution license which allows users to read, copy, distribute and make derivative works, as long as the author of the original work is cited. You may self- archive this article on your own website, an institutional repository or funder’s repository and make it publicly available immediately. 1 23 J Real-Time Image Proc https://doi.org/10.1007/s11554-017-0736-x ORIGINAL RESEARCH PAPER Real-time scene reconstruction and triangle mesh generation using multiple RGB-D cameras 1 2,3 1 Siim Meerits • Vincent Nozick • Hideo Saito Received: 3 April 2017 / Accepted: 26 September 2017 Ó The Author(s) 2017. This article is an open access publication Abstract We present a novel 3D reconstruction system have practical uses in fields such as archeology [42], cin- that can generate a stable triangle mesh using data from ematography [45] and robotics [31]. A majority of the multiple RGB-D sensors in real time for dynamic scenes. works produced so far have focused on static scenes. The first part of the system uses moving least squares However, many interesting applications, such as telepres- (MLS) point set surfaces to smooth and filter point clouds ence [13, 37, 44], require 3D reconstruction in a dynamic acquired from RGB-D sensors. The second part of the environment, i.e. in a scene where geometric and colori- system generates triangle meshes from point clouds. The metric properties are not constant over time. whole pipeline is executed on the GPU and is tailored to For most 3D reconstruction systems, the process can be scale linearly with the size of the input data. Our contri- divided conceptually into three stages: butions include changes to the MLS method for improving 1. Data acquisition Traditionally, acquiring a real-time meshing, a fast triangle mesh generation method and GPU 3D structure of an environment using stereo or multi- implementations of all parts of the pipeline. view stereo algorithms has been a challenging and computationally expensive stage. The advent of con- Keywords 3D reconstruction Á Meshing Á Mesh zippering Á sumer RGB-D sensors has enabled the 3D reconstruc- RGB-D cameras Á GPU tion process to become truly real-time, but RGB-D devices still have their own drawbacks. The generated depth maps tend to be noisy, and the scene coverage is 1 Introduction restricted due to the sensor’s limited focal length, making the following process stages more difficult to 3D surface reconstruction and meshing methods have been achieve. Many reconstruction systems start with a researched for decades in the computer vision and com- preprocessing stage to reduce some noise inherent to puter graphics fields. Its applications are numerous and RGB-D sensors. 2. Surface reconstruction The surface reconstruction & Siim Meerits consolidates available 3D information to a single [email protected] consistent surface. Vincent Nozick 3. Geometry extraction After a surface has been defined, [email protected] it should be converted to a geometric representation Hideo Saito that is useful for a particular application. Some [email protected] commonly used formats are point clouds, triangle meshes and depth maps. 1 Department of Information and Computer Science, Keio University, Yokohama, Japan Our work involves both surface reconstruction and triangle 2 Institut Gaspard Monge, Universite´ Paris-Est Marne-la- mesh generation. We enhance a moving least squares Valle´e, Champs-sur-Marne, France (MLS)-based surface reconstruction method to fit our 3 Japanese-French Laboratory for Informatics, Tokyo, Japan needs. We generate triangle meshes in the geometry 123 J Real-Time Image Proc extraction stage, since they are the most commonly used their ability to integrate noisy input data in real time to representation in computer graphics and are view produce high-quality scene models. However, a major independent. drawback is their high memory consumption. Whelan et al. [50] and Chen et al. [10] propose out-of-core approaches 1.1 Related works where reconstruction volume is moved around in space to lower system memory requirements. However, these We constrain the related literature section to 3D recon- methods cannot capture dynamic scenes. Newcombe et al. struction methods that can work with range image data [40], Dou et al. [16] and Innmann et al. [27] support from RGB-D sensors and provide explicit surface geometry changes to the scene by deforming the reconstructed vol- outputs, such as triangle meshes. As such, view-dependent ume. These methods expect accurate object tracking, which methods are not discussed. can fail under complex or fast movement. Variational volumetric methods, such as those of 1.1.1 Visibility methods Kazhdan et al. [30] and Zach [51], reconstruct surfaces by solving an optimization task under specified constraints. Visual hull methods, as introduced in Laurentini [34], Recently, the method of Kazhdan and Hoppe [29] has reconstruct models using an intersection of object silhou- become a popular choice with Collet et al. [13] further ettes from multiple viewpoints. Polyhedral geometry developing it for use in real-time 3D reconstruction, albeit [20, 39] has become a popular representation of hull utilizing an incredible amount of computational power. structure. To speed up the process, Li et al. [36], Duck- Indeed, the global nature of the optimization comes with a worth and Roberts [17] developed GPU-accelerated great computational cost, making it infeasible in most sit- reconstruction methods. The general drawback with those uations with consumer-grade hardware. approaches is that they need to extract objects from an image background using silhouettes. In a cluttered and 1.1.3 Point-based methods open scene, this is difficult to do. We also may wish to include backgrounds in reconstructions, but this is gener- MLS methods have a long history in data science as a tool ally not supported. for smoothing noisy data. Alexa et al. [2] used this concept Curless and Levoy [15] use a visual hull concept toge- in computer visualization to define point set surfaces (PSS). ther with range images and define a space carving method. These surfaces are implicitly defined and allow points to be Thanks to the range images, background segmentation refined by reprojecting to them. Since then, a wide variety becomes a simple task. However, space carving is designed of methods based on PSS have appeared—see Cheng et al. to operate on top of volumetric methods, which means that [11] for a partial summary. In the classical formulation, issues with volumetric methods also apply here. Using Levin [35] and Alexa et al. [3] approximate local surfaces multiple depth sensors or multiple scans of scenes can around a point as a low-degree polynomial. Alexa and introduce range image misalignments. Zach et al. [52] Adamson [1] simplify the approach by formulating a counters these misalignments by merging scans with a signed distance field of the scene from oriented normals. regularization procedure, albeit with even higher memory To increase result stability, Guennebaud and Gross [22] consumption. When a scene is covered with scans at dif- formulate higher-order surface approximation while ferent scales, fine resolution surface details can be lost. Fleishman et al. [19] and Wang et al. [49] add detail-pre- Fuhrmann and Goesele [21] developed a hierarchical vol- serving MLS methods. Kuster et al. [33] introduce tem- ume approach to retain the maximum amount of details. porally stable MLS for use in dynamic scenes. However, the method is designed to combine a very high Most works to date utilize splatting [53] for visualizing number of viewpoints for off-line processing; as such, it is MLS point clouds. While fast, this approach cannot handle unsuited to real-time use. texturing without blurring, so it is not as well supported in computer graphics as traditional triangle meshes are. It has 1.1.2 Volumetric methods been considered difficult to generate meshes on top of MLS processed point clouds. Regarding MLS, Berger et al. [7] Volumetric reconstruction methods represent 3D data as note that ‘‘it is nontrivial to explicitly construct a contin- grids of voxels. Each volume element can contain space uous representation, for instance an implicit function or a occupancy data [12, 14] or samples of continuous functions triangle mesh’’. Scheidegger et al. [46] and Schreiner et al. [15]. After commodity RGB-D sensors became available, [47] propose advancing front methods to generate triangles the work of Izadi et al. [28] spawned a whole family of 3D on the basis of MLS point clouds.
Load more

Recommended publications
  • Mesh Compression
    Mesh Compression Dissertation der Fakult¨at f¨ur Informatik der Eberhard-Karls-Universit¨at zu T¨ubingen zur Erlangung des Grades eines Doktors der Naturwissenschaften (Dr. rer. nat.) vorgelegt von Dipl.-Inform. Stefan Gumhold aus Tubingen¨ Tubingen¨ 2000 Tag der m¨undlichen Qualifikation: 19.Juli 2000 Dekan: Prof. Dr. Klaus-J¨orn Lange 1. Berichterstatter: Prof. Dr.-Ing. Wolfgang Straßer 2. Berichterstatter: Prof. Jarek Rossignac iii Zusammenfassung Die Kompression von Netzen ist eine weitgef¨acherte Forschungsrichtung mit Anwen- dungen in den verschiedensten Bereichen, wie zum Beispiel im Bereich der Hand- habung extrem großer Modelle, beim Austausch von dreidimensionalem Inhaltuber ¨ das Internet, im elektronischen Handel, als anpassungsf¨ahige Repr¨asentation f¨ur Vo- lumendatens¨atze usw. In dieser Arbeit wird das Verfahren der Cut-Border Machine beschrieben. Die Cut-Border Machine kodiert Netze, indem ein Teilbereich durch das Netz w¨achst (region growing). Kodiert wird die Art und Weise, wie neue Netzele- mente dem wachsenden Teilbereich einverleibt werden. Das Verfahren der Cut-Border Machine kann sowohl auf Dreiecksnetze als auch auf Tetraedernetze angewendet wer- den. Trotz der einfachen Struktur des Verfahrens kann eine sehr hohe Kompression- srate erzielt werden. Im Falle von Tetraedernetzen erreicht die Cut-Border Machine die beste Kompressionsrate von allen bekannten Verfahren. Die einfache Struktur der Cut-Border Machine erm¨oglicht einerseits die Realisierung direkt in Hardware und ist auch als Implementierung in Software extrem schnell. Auf der anderen Seite erlaubt die Einfachheit eine theoretische Analyse des Algorithmus. Gezeigt werden konnte, dass f¨ur ebene Triangulierungen eine leicht modifizierte Version der Cut-Border Machine lineare Laufzeiten in der Zahl der Knoten erzielt und dass die komprimierte Darstellung nur linearen Speicherbedarf ben¨otigt, d.h.
    [Show full text]
  • A Frontal Delaunay Quad Mesh Generator Using the L Norm
    INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING Int. J. Numer. Meth. Engng 2010; 00:1{6 Prepared using nmeauth.cls [Version: 2002/09/18 v2.02] A frontal Delaunay quad mesh generator using the L1 norm J.-F. Remacle1, F. Henrotte1, T. Carrier-Baudouin1, E. B´echet2, E. Marchandise1, C. Geuzaine3 and T. Mouton2 1 Universit´ecatholique de Louvain, Institute of Mechanics, Materials and Civil Engineering (iMMC), B^atimentEuler, Avenue Georges Lema^ıtre 4, 1348 Louvain-la-Neuve, Belgium 2 Universit´ede Li`ege, LTAS, Li`ege,Belgium 3 Universit´ede Li`ege, Department of Electrical Engineering and Computer Science, Montefiore Institute B28, Grande Traverse 10, 4000 Li`ege,Belgium SUMMARY In a recent paper [1], a new indirect method to generate all-quad meshes has been developed. It takes advantage of a well known algorithm of the graph theory, namely the Blossom algorithm, which computes in polynomial time the minimum cost perfect matching in a graph. In this paper, we describe a method that allow to build triangular meshes that are better suited for recombination into quadrangles. This is done by using the infinity norm to compute distances in the meshing process. The alignment of the elements in the frontal Delaunay procedure is controlled by a cross field defined on the domain. Meshes constructed this way have their points aligned with the cross field directions and their triangles are almost right everywhere. Then, recombination with the Blossom-based approach yields quadrilateral meshes of excellent quality. Copyright c 2010 John Wiley & Sons, Ltd. key words: quadrilateral meshing; surface remeshing; graph theory; optimization; perfect matching 1.
    [Show full text]
  • A Laplacian for Nonmanifold Triangle Meshes
    Eurographics Symposium on Geometry Processing 2020 Volume 39 (2020), Number 5 Q. Huang and A. Jacobson (Guest Editors) A Laplacian for Nonmanifold Triangle Meshes Nicholas Sharp and Keenan Crane Carnegie Mellon University Abstract We describe a discrete Laplacian suitable for any triangle mesh, including those that are nonmanifold or nonorientable (with or without boundary). Our Laplacian is a robust drop-in replacement for the usual cotan matrix, and is guaranteed to have nonnegative edge weights on both interior and boundary edges, even for extremely poor-quality meshes. The key idea is to build what we call a “tufted cover” over the input domain, which has nonmanifold vertices but manifold edges. Since all edges are manifold, we can flip to an intrinsic Delaunay triangulation; our Laplacian is then the cotan Laplacian of this new triangulation. This construction also provides a high-quality point cloud Laplacian, via a nonmanifold triangulation of the point set. We validate our Laplacian on a variety of challenging examples (including all models from Thingi10k), and a variety of standard tasks including geodesic distance computation, surface deformation, parameterization, and computing minimal surfaces. CCS Concepts • Mathematics of computing ! Discretization; Partial differential equations; 1. Introduction Discrete Laplacians. For triangle meshes, the de facto standard is the cotan Laplacian (Section 3.3), equivalent to the usual linear finite The Laplacian D measures the degree to which a given function element stiffness matrix. This operator is very sparse, easy to build, u deviates from its mean value in each local neighborhood; it and generally works well for unstructured meshes with irregular hence characterizes a wide variety of phenomena such as the dif- vertex distributions.
    [Show full text]
  • Mesh Models of Images, Their Generation, and Their Application in Image Scaling
    Mesh Models of Images, Their Generation, and Their Application in Image Scaling by Ali Mostafavian B.Sc., Iran University of Science and Technology, 2007 M.Sc., Sharif University of Technology, 2009 A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY in the Department of Electrical and Computer Engineering c Ali Mostafavian, 2019 University of Victoria All rights reserved. This dissertation may not be reproduced in whole or in part, by photocopying or other means, without the permission of the author. ii Mesh Models of Images, Their Generation, and Their Application in Image Scaling by Ali Mostafavian B.Sc., Iran University of Science and Technology, 2007 M.Sc., Sharif University of Technology, 2009 Supervisory Committee Dr. Michael D. Adams, Supervisor (Department of Electrical and Computer Engineering) Dr. Pan Agathoklis, Departmental Member (Department of Electrical and Computer Engineering) Dr. Venkatesh Srinivasan, Outside Member (Department of Computer Science) iii ABSTRACT Triangle-mesh modeling, as one of the approaches for representing images based on nonuniform sampling, has become quite popular and beneficial in many applications. In this thesis, image representation using triangle-mesh models and its application in image scaling are studied. Consequently, two new methods, namely, the SEMMG and MIS methods are proposed, where each solves a different problem. In particular, the SEMMG method is proposed to address the problem of image representation by producing effective mesh models that are used for representing grayscale images, by minimizing squared error. The MIS method is proposed to address the image- scaling problem for grayscale images that are approximately piecewise-smooth, using triangle-mesh models.
    [Show full text]
  • Mesh Generation
    Mesh Generation Pierre Alliez Inria Sophia Antipolis - Mediterranee 2D Delaunay Refinement 2D Triangle Mesh Generation Input: . PSLG C (planar straight line graph) . Domain bounded by edges of C Output: . triangle mesh T of such that . vertices of C are vertices of T . edges of C are union of edges in T . triangles of T inside have controlled size and quality Key Idea . Break bad elements by inserting circumcenters (Voronoi vertices) [Chew, Ruppert, Shewchuk,...] “bad” in terms of size or shape Basic Notions C: PSLG describing the constraints T: Triangulation to be refined Respect of the PSLG . Edges a C are split until constrained subedges are edges of T . Constrained subedges are required to be Gabriel edges . An edge of a triangulation is a Gabriel edge if its smallest circumcirle encloses no vertex of T . An edge e is encroached by point p if the smallest circumcirle of e encloses p. Refinement Algorithm C: PSLG bounding the domain to be meshed. T: Delaunay triangulation of the current set of vertices T|: T Constrained subedges: subedges of edges of C Initialise with T = Delaunay triangulation of vertices of C Refine until no rule apply . Rule 1 if there is an encroached constrained subedge e insert c = midpoint(e) in T (refine-edge) . Rule 2 if there is a bad facet f in T| c = circumcenter(f) if c encroaches a constrained subedge e refine-edge(e). else insert(c) in T 2D Delaunay Refinement PSLG Background Constrained Delaunay Triangulation Delaunay Edge An edge is said to be a Delaunay edge, if it is inscribed in an empty circle
    [Show full text]
  • Mesh Generation for Voxel -Based Objects
    Graduate Theses, Dissertations, and Problem Reports 2005 Mesh generation for voxel -based objects Hanzhou Zhang West Virginia University Follow this and additional works at: https://researchrepository.wvu.edu/etd Recommended Citation Zhang, Hanzhou, "Mesh generation for voxel -based objects" (2005). Graduate Theses, Dissertations, and Problem Reports. 2676. https://researchrepository.wvu.edu/etd/2676 This Dissertation is protected by copyright and/or related rights. It has been brought to you by the The Research Repository @ WVU with permission from the rights-holder(s). You are free to use this Dissertation in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you must obtain permission from the rights-holder(s) directly, unless additional rights are indicated by a Creative Commons license in the record and/ or on the work itself. This Dissertation has been accepted for inclusion in WVU Graduate Theses, Dissertations, and Problem Reports collection by an authorized administrator of The Research Repository @ WVU. For more information, please contact [email protected]. Mesh Generation for Voxel-Based Objects Hanzhou Zhang Dissertation submitted to the College of Engineering and Mineral Resources at West Virginia University in partial ful¯llment of the requirements for the degree of Doctor of Philosophy in Mechanical Engineering Dr. Andrei V. Smirnov, Chair Dr. Ismail Celik Dr. Victor H. Mucino Dr. Ibrahim Yavuz Dr. Frances L.Van Scoy Department of Mechanical and Aerospace Engineering Morgantown, West Virginia 2005 ABSTRACT Mesh Generation for Voxel-Based Objects Hanzhou Zhang A new physically-based approach to unstructured mesh generation via Monte-Carlo simulation is proposed.
    [Show full text]
  • Grouper: a Compact, Streamable Triangle Mesh Data Structure
    IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, VOL. 19, NO. X, XXXXXXX 2013 1 Grouper: A Compact, Streamable Triangle Mesh Data Structure Mark Luffel, Topraj Gurung, Peter Lindstrom, Senior Member, IEEE, and Jarek Rossignac Abstract—We present Grouper: an all-in-one compact file format, random-access data structure, and streamable representation for large triangle meshes. Similarly to the recently published SQuad representation, Grouper represents the geometry and connectivity of a mesh by grouping vertices and triangles into fixed-size records, most of which store two adjacent triangles and a shared vertex. Unlike SQuad, however, Grouper interleaves geometry with connectivity and uses a new connectivity representation to ensure that vertices and triangles can be stored in a coherent order that enables memory-efficient sequential stream processing. We present a linear-time construction algorithm that allows streaming out Grouper meshes using a small memory footprint while preserving the initial ordering of vertices. As a part of this construction, we show how the problem of assigning vertices and triangles to groups reduces to a well-known NP-hard optimization problem, and present a simple yet effective heuristic solution that performs well in practice. Our array-based Grouper representation also doubles as a triangle mesh data structure that allows direct access to vertices and triangles. Storing only about two integer references per triangle—i.e., less than the three vertex references stored with each triangle in a conventional indexed mesh format—Grouper answers both incidence and adjacency queries in amortized constant time. Our compact representation enables data-parallel processing on multicore computers, instant partitioning and fast transmission for distributed processing, as well as efficient out-of-core access.
    [Show full text]
  • R. Garimella, M. Shashkov and P. Knupp, Triangular and Quadrilateral Surface Mesh Quality Optimization Using Local
    Triangular and Quadrilateral Surface Mesh Quality Optimization Using Local Parametrization (LA-UR-02-7190) Rao V. Garimella a, Mikhail J. Shashkov a, Patrick M. Knupp b aMS B284, Los Alamos National Laboratory, Los Alamos, NM 87545. bMS 0847, Sandia National Laboratories, Albuquerque, NM 87185. Abstract A procedure is presented to improve the quality of surface meshes while maintain- ing the essential characteristics of the discrete surface. The surface characteristics are preserved by repositioning mesh vertices in a series of element-based local para- metric spaces such that the vertices remain on the original discrete surface. The movement of the mesh vertices is driven by a non-linear numerical optimization process. Two optimization approaches are described, one which improves the qual- ity of elements as much as possible and the other which improves element quality but also keeps the new mesh as close as possible to the original mesh. Key words: Unstructured surface mesh, triangles, quadrilaterals, numerical optimization, element quality, Jacobian condition number, Reference Jacobian Matrices 1 Introduction Improvement of mesh quality is a very important problem for mesh generation and numerical simulation. The quality of a surface mesh heavily influences the ability of mesh generation algorithms to generate good quality solid meshes. Since surface meshes define external and internal boundaries of computational domains where boundary conditions are imposed, they also influence the ac- curacy of numerical simulations. Email addresses: [email protected] (Rao V. Garimella), [email protected] (Mikhail J. Shashkov), [email protected] (Patrick M. Knupp). Preprint submitted to Elsevier Science 18 November 2002 Many researchers have investigated the issue of improving the quality (element shape and mesh gradation) of triangular, quadrilateral and mixed meshes in the plane [1–6].
    [Show full text]
  • Efficient Construction and Simplification of Delaunay Meshes
    Efficient Construction and Simplification of Delaunay Meshes Yong-Jin Liu∗ Chun-Xu Xu Dian Fan Ying He∗ Tsinghua University Tsinghua University Tsinghua University Nanyang Technological University Input mesh M Delaunay mesh Simplified DM1 Simplified DM2 Exact geodesic Heat method on Heat method on Heat method on Heat method on 7,203 vertices 17,710 vertices 7,203 vertices 3,602 vertices distances M, ǫ = 3.88% DM, ǫ = 1.29% simplified DM1 simplified DM2 Figure 1: We present an efficient algorithm to convert an arbitrary manifold triangle mesh M to a Delaunay mesh (DM), which has the same geometry of M. Our algorithm can also produce progressive Delaunay meshes, allowing a smooth choice of detail levels. Since DMs are represented using conventional mesh data structures, the existing digital geometry processing algorithms can benefit the numerical stability of DM without changing any codes. For example, DMs significantly improve the accuracy of the heat method for computing geodesic distances. Abstract CR Categories: I.3.5 [Computer Graphics]: Computational Ge- ometry and Object Modeling Delaunay meshes (DM) are a special type of triangle mesh where Keywords: Delaunay triangulation, Delaunay mesh, geodesic the local Delaunay condition holds everywhere. We present an Voronoi diagram, Delaunay mesh simplification efficient algorithm to convert an arbitrary manifold triangle mesh M into a Delaunay mesh. We show that the constructed DM has O(Kn) vertices, where n is the number of vertices in M and K 1 Introduction is a model-dependent constant. We also develop a novel algorithm to simplify Delaunay meshes, allowing a smooth choice of detail Delaunay triangulations are widely used in scientific computing in levels.
    [Show full text]
  • Novel Technologies for the Manipulation of Meshes on the CPU and GPU
    Copyright is owned by the Author of the thesis. Permission is given for a copy to be downloaded by an individual for the purpose of research and private study only. The thesis may not be reproduced elsewhere without the permission of the Author. Novel Technologies for the Manipulation of Meshes on the CPU and GPU A thesis presented in partial fulfilment of the requirements for the degree of Masters of Science in Computer Science at Massey University, Palmerston North, New Zealand. Richard John Rountree 2007 i ii Abstract This thesis relates to research and development in the field of 3D mesh data for computer graphics. A review of existing storage and manipulation techniques for mesh data is given followed by a framework for mesh editing. The proposed framework combines complex mesh editing techniques, automatic level of detail generation and mesh compression for storage. These methods work coherently due to the underlying data structure. The problem of storing and manipulating data for 3D models is a highly researched field. Models are usually represented by sparse mesh data which consists of vertex position information, the connectivity information to generate faces from those vertices, surface normal data and texture coordinate information. This sparse data is sent to the graphics hardware for rendering but must be manipulated on the CPU. The proposed framework is based upon geometry images and is designed to store and manipulate the mesh data entirely on the graphics hardware. By utilizing the highly parallel nature of current graphics hardware and new hardware features, new levels of interactivity with large meshes can be gained.
    [Show full text]
  • Generative CSG Models for Real Time Graphics
    () Generative CSG Models for Real Time Graphics B. Guenter & M. Gavriliu Abstract Generative parametric CSG models, introduced by Snyder in 1992, have many desirable properties. Their func- tional representation is generally quite compact and resolution independent, since surfaces are represented as piecewise continuous functional programs. However, in practice they have proven impractical for real time ren- dering because of the difficulty of compactly and exactly representing the implicit curves of intersection between general parametric surfaces and because there was no published algorithm for triangulating the surfaces in real time. Our new algorithm computes an exact, piecewise parametric representation for the implicit curves of inter- section. The new piecewise parametric curve representation is very compact and can be evaluated efficiently at run time, making it possible to change triangulation density dynamically. We have also developed a triangulation algorithm which effectively uses modern GPU’s to render generative CSG models at high speed. Complex genera- tive CSG models made with our system have a memory footprint of just 7-11 KBytes, which is orders of magnitude smaller than the equivalent polygonal mesh representation. 1. Introduction described in Snyder’s seminal work [Sny92a], have many desirable characteristics for real time rendering. They are powerful enough to model many types of objects, especially manufactured things which frequently have simple proce- dural descriptions. The intrinsic generative model descrip- tion is very compact and resolution independent because sur- faces are represented as piecewise continuous functions. Compactness is particularly important because of a long term trend in computer hardware: GPU processor perfor- mance has been increasing at a rate of roughly 71% per year while memory bandwidth has been increasing at only 25% per year [Owe05].
    [Show full text]
  • A Survey of Unstructured Mesh Generation Technology
    A Survey of Unstructured Mesh Generation Technology Steven J. Owen Department of Civil and Environmental Engineering, Carngie Mellon University, Pittsburgh, PA. and ANSYS Inc., Canonsburg, PA. [email protected] Abstract A brief survey of some of the fundamental algorithms in unstructured mesh generation is presented. Included is a discussion and categorization of triangle, tetrahedral, quadrilateral and hexahedral mesh generation methods currently in use in academia and industry. Also included is a brief discussion of smoothing, cleanup and refinement algorithms. An informal survey of currently available mesh generation software is also provided comparing some of their main features. 1. Introduction Automatic unstructured mesh generation is a relatively new field. Within its short life span we have seen tremendous advances in many diverse fields. Once in a while, it is useful to step back from our own expertise and look at the entire picture of what is going on in the field. The purpose of this survey is to give some perspective to what the current trends are in mesh generation and outline some of the major technology areas, who is working in these fields and what software is available. Probably the simplest approach is to first break down the technology based on the shape of element generated. We will consider triangle and quad generation methods in 2D and tetrahedral and hexahedral methods in 3D. Straddled between 2D and 3D, we have surface meshing, which has it’s own set of issues. In addition we have another set of issues dealing with post processing of the mesh including smoothing, cleanup and refinement.
    [Show full text]