meshing
play

Meshing Formally introd by Voronoi in 1908 Meshing Delaunay 1934 - PDF document

Delaunay/Voronoi CS 101 An essential notion CS 101 Meshing Formally introd by Voronoi in 1908 Meshing Delaunay 1934 Applications galore Introduction to Delaunay Triangulations path planning in robotics crystallography


  1. Delaunay/Voronoi CS 101 An essential notion CS 101 Meshing � Formally intro’d by Voronoi in 1908 Meshing � Delaunay 1934 � Applications galore Introduction to Delaunay Triangulations � path planning in robotics � crystallography and Voronoi Diagrams � minimum spanning tree Part I � travel salesman � medial axis CS101 - Meshing 2 All About Proximity Basic Definitions We will see connections with: Let P = (p 1 , p 2 , …, p n ) be n 2D points � nearest neighbors � called sites in this context � by definition Partition the plane in n regions V i � convex hull (CH) � V i = points closer to p i than any other � much less trivial � one of many wonderful properties… = Voro Vorono noi cell ell (caveat: can be unbounded) � circumcircles and triangles � Points with more than one nearest site = Voro Vorono noi dia iagra gram CS101 - Meshing CS101 - Meshing 3 4 Example Example 2 points Points that are equidistant (bisector) circumcircle CS101 - Meshing CS101 - Meshing 5 6

  2. Voronoi cell Voronoi Cell Definition using halfplanes � let be the halfplane: � boundary = bisector of the two sites is in the halfplane � � Then we have: � Consequence: convex cells convex cells CS101 - Meshing CS101 - Meshing 7 8 Voronoi Diagram Example Once we have defined the cell � boundary made out of edges � Voronoi edges � 2 closest sites � Voronoi edges linking vertices � Voronoi vertices � multiple closest sites � remember: cell complex… Four points are cocircular… CS101 - Meshing CS101 - Meshing 9 10 Example Voronoi Example I CS101 - Meshing CS101 - Meshing 11 12

  3. Voronoi Example II Voronoi in Nature? CS101 - Meshing CS101 - Meshing 13 14 Voronoi Properties Some Applications Too many to mention… Motion Planning � cell always convex � what path should a robot follow to stay as far as possible from sites? � if V vertex of degree 3, circumcenter � if p j n.n. of p i they share a V-edge Ambulance Dispatcher � which hospital is closest to scene? � cell unbounded iff site on CH � proof? Farthest point sampling � think about a point at infinity � where to put a new Starbuck’s? � p on CH iff closest point from inf. point CS101 - Meshing CS101 - Meshing 15 16 Delaunay Triangulation (Rough) Examples Definition: � DT is the (straight-line) dual of VD � dual in the sense of: � V cell -> D vertex � V edge -> D edge � V vertex -> D triangle � careful: what happen when 4 cocircular points? An example from the sea… CS101 - Meshing CS101 - Meshing 17 18

  4. Realtime Demo Example in 3D click here CS101 - Meshing CS101 - Meshing 19 20 Properties Proof? Too many… • If ab is in DT, V(a) and V(b) share an edge e. Pick x on e, and trace a circle center on x � “same” as Voronoi through a and b. if there’s a site c in or on the � e.g., if p j n.n. of p i (p i, p j ) is a D-edge circle, x would be in V(c) too! So no site inside. � important property ( Empty Circle ): •Suppose there’s an empty circle through a and b centered on a point x. x is on bisector btw a For every D-edge (p i, p j ), there’s a and b (equidistant). Now wiggle x a bit while circle through p i and p j that doesn’t keeping the circle empty—along bisector. contain ANY other vertex. Clearly, x is on a V-edge, thus ab D-edge. The reverse holds too. CS101 - Meshing CS101 - Meshing 21 22 Applications Numerical Properties Nearest neighbor search: In 2D: maximizes minimum angle � NNG ⊆ DT � stronger: best lexicographic sequence of triangle angles � i.e., if b n.n. of a, then ab is in DT � great for numerics (as detailed later) Medial Axis – generalization (grassfire) � Edelsbrunner ’87 Minimum Spanning Tree � ain’t true in 3D…. � MST is a subgraph of DT � suppose (a,b) edge of MST, and assume (a.b) NOT Delaunay; prove contradiction (use empty circle…) CS101 - Meshing CS101 - Meshing 23 24

  5. Connection to CH Lifting Map We already saw link Lift points (x, y) to paraboloid (x, y, x 2 + y 2 ) � placement of origin irrelevant � CH of sites goes along D-edges DT = CH of 3D points But what if we think higher dim… � I.e., 2D plane and sites are projections of something 3D � great property, valid in nD CS101 - Meshing CS101 - Meshing 25 26 Proof Same for Voronoi? Equation of tangent plane at (a,b): Almost… VD = proj. of tangent envelope Shift upwards by r 2 : Intersection with paraboloid? Before… Now, imagine you have a face (p1,p2,p3) of 3D CH � other vertices are at least r 2 away, right? � thus, they project OUTSIDE of the projected circle � bingo: empty circle, thus the face maps to D-triangle. CS101 - Meshing CS101 - Meshing 27 28

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend