Design and Implementation - Segmentation of Hand Bone for Bone Age Assessment - page 72

(D) The Assignment of Pixel's Intensity Value To Cluster Center

1, argmin j = 1 D ( T )

ij

(3.43)

w ji =

0, Otherwise

(E) Cluster Center Recalculation

w ji x i

n
i

C j

= (3.44)

n
i

w ji

(F) Stopping Criteria

Iterate until E(T) is less than tolerant error value '∂'

C j − C T − 1

j

E ( T ) = ≤ ∂

(3.45)

(G) Iterations

Repeats with T ← T + 1 in step 2 if stopping criteria is not achieved

(H) Cluster Center Replacement Fill the pixels belong to cluster with lower
value of cluster center with zero intensity and remain the pixels intensity that
belong to cluster with higher value of cluster center.

0, argmin j = 1 Cj
x i , otherwise

x ( j )
i

=

(3.46)

(I) Cluster Center Replacement
(ii) Repeat the process with k = 2
(ii) Repeat the process with k = 3

(J) Cluster Center Replacement

Fill the pixels belong to clusters with lower value of cluster center with zero
intensity and remain the pixels intensity that belongs to cluster with highest value
of cluster center.

x i , argmax j = 1 Cj

x ( j )
i

=

0, otherwise
0, otherwise

(3.47)

Grey Level Co-occurrence Matrix (GLCM) proposed by Haralick et al. [ 19 ]
suggested a way to describe texture of image by statistical measurement. A GLCM
is a matrix that contains the probability of gray scale i occurs together with the
gray level j at a specific distance, in a specific direction. The distance and direction