The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part B3b. Beijing 2008
determinate the number of clusters for a clustering. Another
issue is how to calculate the fuzzy c-partition matrix. In this
paper the number of cluster is determined by user a prior. The
second issue is solved by calculating colour similarity
measurement as follows.
For a given vector set V = {v ls ... , v„), the fuzzy c-partition
matrix can be calculated as follows.
YjP(u k ,Vj) m -'
k=1
simplest is called mutation. Just as mutation in living things
changes one gene to another, so mutation in a genetic algorithm
causes small alterations at single points in an individual's code.
The second method is called crossover, and entails choosing
two individuals to swap segments of their code, producing
artificial offspring that are combinations of their parents. This
process is intended to simulate the analogous process of
recombination that occurs to chromosomes during sexual
reproduction. Common forms of crossover include single-point
crossover, in which a point of exchange is set at a random
location in the two individuals' genomes, and one individual
contributes all its code from before that point and the other
contributes all its code from after that point to produce an
offspring.
where m e (1, go) is the weighting exponent on each fuzzy
membership. The larger m is the fuzzier the partition is, «,• is
central vector for cluster /, and p{u h vj) is a similarity measure
between vectors «, and Vj and can be calculated by
M(u n v ] ) = exp(-^£/ («,., Vj)) cos(£ 2 #(ii,. , v y )) (4)
where k\ and k 2 are parameters and d{w„ vj) and 6{u n vj) are the
distance and the angle between u, and Vj as follows,
( L
d(u n v.) =
V
(5)
2.3 Colour Histogram
Colour histogram is an important technique in colour image
analysis, because of its efficiency, effectiveness and triviality in
computation (Pratt, 1991). Generally speaking, a colour
histogram represents the statistical distribution of the colours in
a colour image on all colours in a colour space.
Given a colour space divided into / colour bins, the colour
histogram of the colour image with n pixels is represented as a
vector H = [h 0 , ..., hi.i], in which each entry hj indicates the
statistical figures of the colours in the colour image which
belong to the ;th bin, i.e.,
(
#(«,., v.) = arccos
v
Yj U » V J>
1=1
(6).
h.=-i = 0,l,,I-l (7)
n
where is the number of pixels with colours in the ith colour
bin.
2.2 Genetic Algorithm
Genetic Algorithms (GA) are computer procedures that employ
the mechanics of natural selection and natural genetics to
evolve solutions to problems (Goldberg, 1989). Given a specific
problem to solve, the input to the GA is a set of potential
solutions to that problem encoded in some fashion, and a metric
called a fitness function that allows each candidate to be
quantitatively evaluated. These candidates may be solutions
already known to work, with the aim of the GA being to
improve them, but more often they are generated at random. In
a pool of randomly generated candidates, these promising
candidates are kept and allowed to reproduce by evaluating
each candidate according to the fitness function using a series
of genetic operations: selection, crossover and mutation.
There are many different techniques by which a genetic
algorithm can be used to select the individuals to be copied over
into the next generation. Roulette-wheel selection is one of the
most common methods. It is a form of fitness proportionate
selection in which the chance of an individual's being selected
is proportional to the amount by which its fitness is greater or
less than its competitors' fitness. Conceptually, this can be
represented as a game of roulette, each individual gets a slice of
the wheel, but more fit ones get larger slices than less fit ones.
The wheel is then spun, and whichever individual owns the
section on which it lands each time is chosen. Once selection
has chosen fit individuals, they must be randomly altered in
hopes of improving their fitness for the next generation. There
are two basic strategies to accomplish this. The first and
Let RGB colour space be discretized along the R, G, and B axes
by the numbers N R , N G , and N B , respectively. Then the total N
(= N r x Nq x Nj) bins are available. These bins are coded in
such a sequence from R to G and then from G to B. According
to the specified discretizing and coding scheme the index of
each bin can be represented as
i - R + N g xG + N B 2 x B (8)
where R = 0, 1, ... ,N R -1, G = 0, 1, ...,N G -1, and 5 = 0, 1,...,
N B -1.
Then the pixel (r p , g p , b p ) will be in the bin with the index i p ,
r N
r p 1 v R
+ N g x
S p N g
+ N B x
b p N t
[ 256 J
[_ 256
256 _
J is an integral operator.
3. DESCRIPTION OF PROPOSED APPROACH
3.1 Fuzzy Segmentation
Based on the concept of the fuzzy c-partition above mentioned,
the colour segmentation approach is designed. The approach
consists of three steps: (1) Pre-clustering. This process includes
finding an initial centre vector set U 0 and indicating the ranges
in which the centre vectors are chosen in the following optimal
procedure. This procedure is finished by using a histogram-