Digital Signal Processing Reference
In-Depth Information
Fig. 2.21
Signal flow diagram of proposed SEFCM algorithm
Fig. 2.22
Signal flow diagram of proposed CEFCM algorithm
together, the CEFCM adopts three dimensional eigen-subspaces data for classifica-
tion. The function block diagram of the CEFCM is shown in Fig.
2.22
. Similar to
SEFCM, we also construct new covariance matrices that are similar to (
2.47
)by
using the eigen-subspace data
z
q
in stead of
x
q
in color images. In view of statistical
inference and fuzzy property, we can construct a new covariance matrix for the
j
th
cluster center as
N
q
=
1
u
jq
z
q
z
q
.
1
C
z
j
=
(2.53)
q
1
u
jq
∑
=
It is not necessary to iteratively rebuild the covariance matrix and construct the
new eigen-subspaces from the original color images because the color objects are
selected under our inspection. We can gradually adjust the direction of principal axes
by using the already built eigen-subspaces so that large amount of transformation
computations can be saved. In updating procedures, we adopt the new covariance
