Geometric Data Structures: G H e D I G H e D I

Geometric Data Structures
Organization of points, lines, planes, to support faster processing Applications

Astrophysical simulation evolution of galaxies Graphics computing object intersections Data compression Points are representatives of 2x2 blocks in an image Nearest neighbor search
K-D Trees
CSE 373 Data Structures Lecture 22
12/6/02
K-D Trees - Lecture 22
k-d Trees
Jon Bentley, 1975 Tree used to store spatial data.
Nearest neighbor search. Range queries. Fast look-up
y d b c x i g e
Range Queries
i h f y d b c x Circular query
K-D Trees - Lecture 22 4
g e
h f
k-d tree are guaranteed log2 n depth where n is the number of points in the set.
Traditionally, k-d trees store points in ddimensional space which are equivalent to vectors in d-dimensional space.
12/6/02 K-D Trees - Lecture 22 3
Rectangular query
12/6/02
Nearest Neighbor Search

i g y d b c x Nearest neighbor is e.
12/6/02 K-D Trees - Lecture 22 5 12/6/02
k-d Tree Construction

If there is just one point, form a leaf with that point. Otherwise, divide the points in half by a line perpendicular to one of the axes. Recursively construct k-d trees for the two sets of points. Division strategies
divide points perpendicular to the axis with widest spread. divide in a round-robin fashion (book does it this way)
h e query f
k-d Tree Construction (1)

i g y d b c x divide perpendicular to the widest spread.
12/6/02 K-D Trees - Lecture 22 7 12/6/02

i x s1 h e f
h e f y d
b c
s1

i g y
s2

i g y
s2
x s1 h y s2 f d b
x s1 h y s2 f x s3
d b
c
s1
a
s3
c
s1
12/6/02
12/6/02
10

i g y
s2

i g y
s2
x s1 h y s2 f x s3 a d b
x s1 h y s2 f x s3 a b
d b
a
s3
c
s1
a
s3
c
s1
12/6/02
11
12/6/02
12

i g y
s4

i g y
s4
x s1 h y s2 f x s3 a b y s4 d
s5 s2
x s1 h y s2 f x s3 a b x s5 y s4
d b
s2
a
s3
c
s1
a
s3
b c
s1
12/6/02
13
12/6/02
14

i g y
s4

i g y
s4
x s1 h y s2 f x s3 a b x s5 d y s4 d
s5 s2
x s1 h y s2 f x s3 a b x s5 d e y s4
d
s5
s2
a
s3
b c
s1
a
s3
b c
s1
12/6/02
15
12/6/02
16

i g y
s4

i g y
s4
x s1 h y s2 f x s3 a b x s5 d e y s4 g d
s5 s2
x s1 h y s2 f x s3 a b x s5 d e y s4 g y s6
d
s5
s6
s2
a
s3
b c
s1
a
s3
b c
s1
12/6/02
17
12/6/02
18

i g y
s4

i y s6 g y
s4
x s1 h y s2 f x s3 a b x s5 d e y s4 g y s7
x s1 h y s2 f x s3 a b x s5 d e y s4 g c y s7 y s6
d
s5
s6
d
s5
s6
s2
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
19
12/6/02
20

i g y
s4

i y s6 g y
s4 s8
x s1 h y s2 f x s3 a b x s5 d e y s4 g c y s7 f
x s1 h f x s3 a b x s5 d e y s2 y s4 g c y s7 f y s6 y s8
d
s5
s6
d
s5
s6
s2
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
21
12/6/02
22

i g y
s4 s8

k-d tree cell i y s6 g y
s4 s8
x s1 h f x s3 a b x s5 d e y s2 y s4 g c y s7 f h
x s1 h f x s3 a b x s5 d e y s2 y s4 g c y s7 f h y s6 y s8 i
d
s5
s6
s2
y s8
d
s5
s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
23
12/6/02
24
2-d Tree Decomposition

2
k-d Tree Splitting

i g y d b c x e h f sorted points in each dimension 1 2 3 4 5 6 7 8 9 x a d g be i c h f y a c b d f e h g i max spread is the max of fx -ax and iy - ay. In the selected dimension the middle point in the list splits the data. To build the sorted lists for the other dimensions scan the sorted list adding each point to one of two sorted lists.
K-D Trees - Lecture 22 26
3
12/6/02 K-D Trees - Lecture 22 25 12/6/02
k-d Tree Splitting

i g y d b c x y a b d eg c f h i
k-d Tree Construction Complexity

First sort the points in each dimension.
O(dn log n) time and dn storage. These are stored in A[1..d,1..n]
sorted points in each dimension 1 2 3 4 5 6 7 8 9 x a d g be i c h f y a c b d f e h g i h indicator for each set f a b c de f g h i 0 0 1 00 1 0 1 1 scan sorted points in y dimension and add to correct set
Finding the widest spread and equally divide into two subsets can be done in O(dn) time. We have the recurrence
T(n,d) < 2T(n/2,d) + O(dn)
Constructing the k-d tree can be done in O(dn log n) and dn storage
Node Structure for k-d Trees

A node has 5 fields
axis (splitting axis) value (splitting value) left (left subtree) right (right subtree) point (holds a point if left and right children are null)
k-d Tree Nearest Neighbor Search

Search recursively to find the point in the same cell as the query. On the return search each subtree where a closer point than the one you already know about might be found.
12/6/02
29
12/6/02
30
k-d Tree NNS (1)

query point i g y
s4 s8
k-d Tree NNS (2)

query point x s1 i y s6 g y
s4 s8
h f x s3 a b
y s2 y s4 x s5 d e g c y s7 f
d
s5
s6
s2
y s8 h i
d
s5
s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
31
12/6/02
32
k-d Tree NNS (3)

query point i g y
s4 s8
k-d Tree NNS (4)

s4 w s8 s6
h f x s3 a b
d
s5
s6
s2
y s8 h i
d
s5
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
33
12/6/02
34
k-d Tree NNS (5)

query point i g y
s4 w s8 s6
k-d Tree NNS (6)

s4 w s8 s6
h f x s3 a b
d
s5
s2
y s8 h i
d
s5
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
35
12/6/02
36
k-d Tree NNS (7)

query point i g y
s4 s8
k-d Tree NNS (8)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
37
12/6/02
38
k-d Tree NNS (9)

query point i g y
s4 s8
k-d Tree NNS (10)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
39
12/6/02
40
k-d Tree NNS (11)

query point i g y
s4 s8
k-d Tree NNS (12)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
41
12/6/02
42
k-d Tree NNS (13)

query point i g y
s4 s8
k-d Tree NNS (14)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
43
12/6/02
44
k-d Tree NNS (15)

query point i g y
s4 s8
k-d Tree NNS (16)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
45
12/6/02
46
k-d Tree NNS (17)

query point i g y
s4 s8
k-d Tree NNS (18)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
47
12/6/02
48
k-d Tree NNS (19)

query point i g y
s4 s8
k-d Tree NNS (20)

s4 s8
h f x s3 a b
d
s5
w s6
s2
y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
a
s3
b
s1
s7
c x
12/6/02
49
12/6/02
50
k-d Tree NNS (21)

query point i g y
s4 s8
Main is NNS(q,root,null,infinity)
Nearest Neighbor Search

x s1
NNS(q: point, n: node, p: point, w: distance) : point { if n.left = null then {leaf case} if distance(q,n.point) < w then return n.point else return p; else if w = infinity then if q(n.axis) < n.value then p := NNS(q,n.left,p,w); w := distance(p,q); if q(n.axis) + w > n.value then p := NNS(q, n.right, p, w); else p := NNS(q,n.right,p,w); w := distance(p,q); if q(n.axis) - w < n.value then p := NNS(q, n.left, p, w); else //w is finite// if q(n.axis) - w < n.value then p := NNS(q, n.left, p, w); w := distance(p,q); if q(n.axis) + w > n.value then p := NNS(q, n.right, p, w); return p }
h f x s3 a b
y s6 y s8 h i
d
s5
w s6
s2
a
s3
b
s1
s7
c x
12/6/02
51
The Conditional
q(n.axis) + w > n.value
Notes on k-d NNS

Has been shown to run in O(log n) average time per search in a reasonable model. (Assume d a constant) Storage for the k-d tree is O(n). Preprocessing time is O(n log n) assuming d is a constant.
n.axis = x
q w Current nearest point n.value q(n.axis) + w
12/6/02
53
12/6/02
54
Geometric Data Structures

Geometric data structures are common. The k-d tree is one of the simplest.
Nearest neighbor search Range queries
Other data structures used for

3-d graphics models Physical simulations
12/6/02
55
10

Geometric Data Structures: G H e D I G H e D I

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Geometric Data Structures: G H e D I G H e D I

Hochgeladen von

Copyright:

Verfügbare Formate

Geometric Data Structures

Organization of points, lines, planes, to support faster processing Applications

K-D Trees - Lecture 22

Nearest Neighbor Search

k-d Tree Construction

K-D Trees - Lecture 22

k-d Tree Construction (1)

k-d Tree Construction (2)

K-D Trees - Lecture 22

k-d Tree Construction (3)

k-d Tree Construction (4)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (5)

k-d Tree Construction (6)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (7)

k-d Tree Construction (8)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (9)

k-d Tree Construction (10)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (11)

k-d Tree Construction (12)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (13)

k-d Tree Construction (14)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (15)

k-d Tree Construction (16)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree Construction (17)

k-d Tree Construction (18)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

2-d Tree Decomposition

k-d Tree Splitting

k-d Tree Splitting

k-d Tree Construction Complexity

Node Structure for k-d Trees

k-d Tree Nearest Neighbor Search

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree NNS (1)

k-d Tree NNS (2)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree NNS (3)

k-d Tree NNS (4)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree NNS (5)

k-d Tree NNS (6)

K-D Trees - Lecture 22

K-D Trees - Lecture 22

k-d Tree NNS (7)