Professional Documents
Culture Documents
1, 2
(Abstract) In this article we study the digital cubical homology groups of digital images which are based on the cubical
homology groups of topological spaces in algebraic topology. We investigate some fundamental properties of cubical homology
groups of digital images. We also calculate cubical homology groups of certain 2-dimensional and 3-dimensional digital images.
We give a relation between digital simplicial homology groups and digital cubical homology groups. Moreover we show that the
Mayer-Vietoris Theorem need not be hold in digital images.
Keywords: Digital Topology; Digital Cubical Set; Digital Cubical Homology Group; Euler Characteristic; Mayer-Vietoris
Theorem
1. INTRODUCTION
Topological invariants are useful in many applications related
to digital imaging and geometric modeling, and homology is a
classical one. The higher homotopy groups and homology
groups are useful algebraic tools in a large number of
topological problems, and are computational tools of
algebraic topology. The digital fundamental group is a nice
tool to classify the digital images with k-adjacency relations
but it does not yield completely information in a great class of
explicit problems. We need to establish a new algebraic
scructure which is called the digital homology groups in order
to classify the various digital images with k-adjacency
relations. The digital cubical homology can be an important
tool to classify digital images. Many researchers (Kaczynski,
Mischaikow, Mrozek, Allili, Tannenbaum, Kalies, Watson,
Pilarczyk, Zelazna and Kot) have studied cubical homology.
Kalies, Mischaikow and Watson [8] introduce a method for
computing the homology groups of cellular complexes
composed of cubes. The algorithm used in the homology
computations is based on a local reduction procedure, and
they give an estimate of its computational complexity. This
estimate is rigorous in two dimensions, and they conjecture its
validity in higher dimensions.
Allili, Mischaikow and Tannenbaum [1] combine a new
method combinatorial topology to compute the number of
connected components and holes of objects in a given image,
and fast segmentation methods to extract the objects. Their
computational method for determining the homology groups
is based on a reduction process of the size of the chain
complex by local simplification in such a way that the
homology is preserved at each step.
Kaczynski, Mischaikow and Mrozek [7] present cubical
sets and the algebra of cubical sets. Also, they define cubical
homology and investigate its most elementary properties.
They present a computational approach to homology with the
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 160 -
2. PRELIMINARIES
n
2.1
[3]
For
positive
integer
with
IJICS
2
[a, b] Z = {z Z a z b}
(2.1)
(2.2)
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 161 -
IJICS
[0,1] [2] [4,5], [0,1] [3] [4,5], [0,1] [2,3] [4], [0,1] [2,3] [5]},
(3.1)
[ ]Z = [a, a ]Z
then we write a
Q = I 1 I 2 I n Z n
(3.2)
[1] [2] [5], [0] [3] [5], [1] [2] [4], [1] [3] [4],
( ) and
( ).
those of dimension q by K q Z
and
K q = Q : Q K q
(3.5)
n
K = K q
( X , k ) . More
(3.6)
q =0
, Q ,...Q K of
Given any finite collection Q
1
2
n
q
q-dimensional elementary digital chains, we can consider
sums of the form
c = 1Q 1 + 2 Q 2 + ... + n Q n
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 162 -
(3.7)
where
IJICS
0, Q = [a ]
q Q =
[a + 1] [a ], Q = [a, a + 1]
()
i = 0, , then c
-adjacency is denoted
Q K q , define
Q = IP .Define
()
dim I
q Q = q1 IP + ( 1) I q2 P
Q : K qn by
1, P = Q
Q (P ) =
0, P Q
(3.8)
c = 1Q 1 + 2 Q 2 + ... + m Q m , then
q (c ) = 1 q Q 1 + 2 q Q 2 + ... + m q Q m
chains of Z with
K q and K q .
dim c1
q : C q ( X ) C q1 ( X )
(3.9)
Let (X, )
q C q ( X ) C q1 ( X ) .
c 2 C q22 by
4.
c1 , P c 2 , Q P Q
(3.10)
q2
q : C q C q 1
(3.11)
(3.16)
,Q
(3.15)
P Q = P Q
q1
c1 q c 2
P K q1 and Q K q2 , set
q c1 c 2 = q c1 c 2 + ( 1)
(3.14)
(3.13)
c1 c 2 =
(3.12)
CUBICAL HOMOLOGY IN
IMAGES
DIGITAL
-adjacency.
q : C q ( X ) C q1 ( X )
(4.1)
q +1 : C q+1 ( X ) C q ( X )
(4.2)
and denoted by Bq ( X ) .
Since Proposition 3.7, each digitally cubical q-boundary of
digitally cubical (q+1)-chains is again a digitally cubical
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 163 -
IJICS
H q ( X ) = Z q ( X ) Bq ( X ).
The homology of X is the collection of all homology groups
of X, that is,
H ( X ) = H q ( X ) q
Given z Z q ( X ) ,
[z ] H q ( X )
(4.3)
of z in X.
To simplification, in 0 ( X ,4 )
, k = 0
Cq (X ) =
0, k 0
(4.5)
0
2
1
C14 ( X )
C 04 ( X )
0
0
C1 ( X ) = 0, B0 ( X ) = 0 , and therefore, H 0 ( X ) .
Since C q ( X ) = 0 for all q 1 , H q ( X ) = 0 for all
Furthermore Z 0 ( X ) C 0 ( X ) = . Since
()
(b ) = a
1 b1 = a1 + a 2
1
q 1 . Therefore,
(4.7)
+ a 3
( )
(b ) = a
1 b2 = a 3 + a 4
1
+ a 4
(4.8)
, k = 0
H q ( X ) =
0, k 0.
(4.6)
()
( )
+ (b )+ (b )
)(b ) + ( + )(b )
1 1b1 + 2 b2 + 3 b3 + 4 b4 = 1 1 b1 + 2 1 b2
3 1
()
( )
= ( 1 3 ) b1 + (1
+ ( + ) b .
2
4 1
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 164 -
1 = 2 = 3 = 4 .
Hence
{(
4
1
(X ) = Z (X ) .
4
1
We turn to computing H
solution to the equation
4
0
()
1 b1 = a1 + a 2
1 b1 + b4
Thus,
( )
b ) = a + a .
1
, q = 0,2
H q6 MSS 6' =
0, q 0,2.
1 b1 + b4 = a1 + a 4
2
That is,
is a digital cubical set with 6-adjacency (see figure 1), then its
digital cubical homology groups are
1 1b1 + 2 b2 + 3 b3 + 4 b4 = b1 .
This implies that
, k = 0,1
H k4 ( X ) =
0, k 0,1.
MSS 6' = {c 0 = (0,0,0), c1 = (1,0,0), c 2 = (1,1,0), c 3 = (0,1,0 ),
14 ( X ) = b1 b2 b3 b4 .
4
that
Theorem 4.5 If
IJICS
{a1 a 2 , a1 a 4 , a1 a 3 } B04 ( X ) .
a1 a 2 a 3 a 4 .
Now consider an arbitrary chain
z C 04 ( X ) .Then
z = 1 a1 + 2 a 2 + 3 a 3 + 4 a 4 .
'
[z ]X = [ 1a1 + 2 a 2 + 3 a 3 + 4 a 4 ]X
'
2 (MSS 6' ,6) = {[0,1] [0,1] [0], [0,1] [0] [0,1], [0,1] [1] [0,1]
= 1 [a1 ]X + 2 [a 2 ]X + 3 [a3 ]X + 4 [a 4 ]X
= ( 1 + 2 + 3 + 4 )[a1 ]X
(4.9)
where the last equality comes from that fact that all the
elementary chains are homologous. Therefore, we can think
of every element of H q ( X ) = Z q ( X ) Bq ( X ) as being
[1] [0,1] [0], [1] [0] [0,1], [1] [1] [0,1], [0,1] [1] [0],
[0] [1] [0,1], [0,1] [0] [1], [0] [0,1] [1], [1] [0,1] [1],
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 165 -
IJICS
c 0 = [0][0][0], c1 = [1][0][0],
0 (MSS ,6) = {[0] [0] [0], [1] [0] [0], [1] [1] [0],
c 2 = [1][1][0], c3 = [0][1][0],
[0] [1] [0], [0] [0] [1], [1] [0] [1], [1] [1] [1],
c 4 = [0][0][1], c5 = [1][0][1],
'
6
c 6 = [1][1][1], c 7 = [0][1][1].
) {
3
2
1
0
C 26 MSS 6'
C16 MSS 6'
0
C06 MSS 6'
0
(4.10)
2 ( e0 ) = d3 + d6 d4 d0
2 (e1 ) = d 0 + d1 d8 d 4
2 (e2 ) = d 6 + d 7 d11 d5
in 1 MSS 6' ,6
2 (e4 ) = d 2 + d 7 d9 d1
in 0 MSS 6' ,6
2 (e3 ) = d3 + d5 d10 d 4
'
2 ( a 0 e0 + a1 e1 + a 2 e 2 + a 3 e3 + + a 4 e 4 + a 5 e5 ) =
( a 0 + a1 )d 0 + (a1 a 4 )d1 + ( a 0 + a 4 )d 2 + (a 0 + a 3 )d 3
+ ( a a )d + (a + a )d + (a + a )d
1
+ (a 2 + a 4 )d 7 + (a1 a 5 )d8
+ (a 4 a 5 )d 9 + (a 3 + a 5 )d10 + (a 2 + a 5 )d11 .
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 166 -
IJICS
a 4 = a 0 a3
a 2 = a 0 a1
(a 0 + a1 )d 0 + (a1 a 4 )d1 + (a 0 + a 4 )d 2 + (a 0 + a 3 )d 3
+ (a a )d + ( a + a )d + (a + a )d
a 7 = a 0 a1 + a3 a5
+ (a 2 + a 4 )d 7 + ( a1 a 5 )d8
+ ( a 4 a 5 )d 9 + ( a 3 + a 5 )d10 + ( a 2 + a 5 )d11 = 0
a10 = a 0 a1 + a3 a5 + a8
Hence we get
a11 = a5 a10 = a 0 + a3 a5 a8 .
Hence we get
a 0 = a1 = a 2 = a3 = a 4 = a5 .
'
a9 = a1 a8
we have
a6 = a3 + a5
'
Z 16 ( MSS 6' ) = {a 0 d 0 + a1 d1 + (a 0 a1 )d 2 + a3 d3 +
(a 0 a3 )d 4 + a5 d5 + ( a3 + a5 )d 6 + (a 0 a1 + a3 a5 )d 7
+ a d + (a a )d + (a a + a a + a )d +
8
10
(a 0 + a3 a5 a8 )d11 : ai Z } Z 5 .
'
we have
1 (d0 ) = c4 c0
1 (d 7 ) = c7 c3
1 (d 2 ) = c3 c0
1 (d3 ) = c7 c4
1 (d 4 ) = c5 c1
1 (d5 ) = c6 c2
1 (d 6 ) = c2 c1
1 (d 7 ) = c6 c5
1 (d8 ) = c1 c0
1 (d9 ) = c5 c4
1 (d10 ) = c2 c3
'
'
1 (a 0 d 0 + a1 d1 + a 2 d 2 + a 3 d 3 + + a 4 d 4 + a 5 d 5 +
a d + a d + a d + a d + a d + a d ) =
'
10
10
: k i Z , i = 0,1,2,3,4} Z 5 .
1 (d11 ) = c6 c7
11 11
( a 0 a1 a 2 )c 0 + (a 0 a 3 a 4 )c1 + (a 3 a 5 + a 6 )c 2
+ (a 2 a 6 a 7 )c3 + (a1 a8 a10 )c 4 + (a 4 a 9 + a11 )c5
+ (a 5 + a 9 + a10 )c6 + (a 7 + a8 a10 )c7
(4.11)
Solving the equation
'
u 0 = a 0 c0 + a1c1 + a 2 c2 + a3 c3 + a 4 c4 + a5 c5 +
a 6 c6 + a 7 c7
can be written as
+ (a5 + a6 )d5 + a6 d6 + a7 d7 ) + ai c0 .
IJICS
'
i =0
(a 0 + a1 + a 2 + a3 + a 4 + a5 + a 6 + a 7 )c0 .
Therefore, H 0 ( MSS 6 ) Z . Thus we have the required
6
'
result:
Z , q = 0,2
H q6 ( MSS 6' ) =
0, q 0,2.
'
X such that
that
n values
Z , q = 0
H q6 ( MSS 6' ) = Z 5 , q = 1
0, q 0,1
H n (X1 X 2 ) H n (X1) H n (X 2 )
(5.1)
H n ( X ) H n 1 ( X 1 X 2 )
be a digital image
Z , n = 0,1
H n4 ( X ) =
0, n 0,1.
p = rankH pk ( X )
(4.12)
(5.2)
Let
( X , k ) = (1) p .
p
p =0
6
0
'
6
6
1
'
6
Z , n = 0
H n4 ( X 1 X 2 ) = H n4 ( X 1 ) = H n4 ( X 2 ) =
0, n 0
(5.3)
Moreover, H 0 ( X ) = Z since
4
X is 4-connected digital
IJICS Volume 1, Issue 7 October 2012 PP. 178-187 www.iji-cs.org Science and Engineering Publishing Company
- 168 -
(5.4)
q 2.
k:Z Z Z
c k (c ) = (c, c )
(5.5)
j
k
0 H 14 ( X )
Z Z Z 0
H 14 ( X ) / Kerj Im j
IJICS
(5.6)
Im j = 0. As
Author Introduction
Ismet Karaca- was born in Afyon,
Turkey on January 5th, 1969. He received a
4
1
6. CONCLUSION
REFERENCES
[1]
topology.
173176.
[2]
[3]
[4]