Slide 1

By Dor Lahav

Slide 2

Overview Straight Skeletons Convex Polygons Constrained Voronoi diagrams and Delauney triangulations

Slide 3

Straight Skeletons

Slide 4

Straight Skeleton- Motivation The occurrence of curved edges in the line segment Voronoi diagram V(G) is a disadvantage in the computer representation. a different alternative to V(G)- the Straight Skeleton.

Slide 5

Straight Skeleton

Slide 6

A few definitions: - A connected component of G is called a figure. - Every simple polygon which rises from a figure, is called a wavefront.

Slide 7

Straight Skeleton There are two types of changes: - Edge event: a wavefront edge collapses to length zero. - Split event: a wavefront edge splits due to interference or self interference. - After each type of event, we have a new set of wavefronts.

Slide 8

Straight Skeleton The edges of S(G) are pieces of angular bisectors traced out by wavefront vertices. Each vertex of S(G) corresponds to event. So we get a unique structure defining a polygonal partition of a plane.

Slide 9

Straight Skeleton Lets see an example

Slide 10

Slide 11

Straight Skeleton How many faces in the diagram? Each segment of G gives rise to two wavefront edges and thus to two faces, one on each side of the segment. Each terminal of G gives rise to one face. This gives a total of 2m+t = O(n) faces(m-edges, t- terminals). There is also an exact bound on the number of vertices

Slide 12

Straight Skeleton Lemma: Let G be a planar straight graph on n points, t of which are terminals. The number of(finite and infinite) vertices of S(G) is exactly 2n+t-2.

Slide 13

Straight Skeleton The Straight Skeleton has a 3- dimensional interpretation obtained by defining the height of a point x in the plane as the unique time when x is reached by the wavefront. Thats how S(G) lifts out a polygonal surface. Points of G have height zero.

Slide 14

Straight Skeleton

Slide 15

Convex Polygons

Slide 16

Let C be a convex n-gon in the plane. The medial axis M(C) of C is a tree whose edges are pieces of angular bisectors of Cs sides. There is a simple randomized incremental algorithm that computes M(C) in O(n) time.

Slide 17

Convex Polygons

Slide 18

Slide 19

Slide 20

Slide 21

Slide 22

Slide 23

Slide 24

Slide 25

Slide 26

Slide 27

Slide 28

Done!

Slide 29

Running time? M(C) has an upper bound of 2n-3 edges. Each edge belongs to two faces. Hence, the average number of edges of a face is Thats a constant time per face. So, a total running time of O(n).

Slide 30

Convex Polygons- Voronoi diagram

Slide 31

Slide 32

Slide 33

Slide 34

Slide 35

Slide 36

Slide 37

Slide 38

Constrained Voronoi diagrams and Delauney triangulations

Slide 39

Constrained Voronoi diagrams Let S be a set of n point sites in the plane. Let L be a set of non-crossing line segments spanned by S. We define the bounded distance between two points x and y in the plane as:

Slide 40

Constrained Voronoi diagrams For each segment l, the regions clipped by l from the right are extended to the left of l, as if only the sites of these regions were present. The regions clipped by l from the left are extended similarly.

Slide 41

Constrained Delauney triangulations If we dualize now by connecting sites of regions that share an edge, a full triangulation that includes L is obtained.

Slide 42