Information Technology Reference
In-Depth Information
complex feature vector form that aims at obtaining a higher recognition effect
and a faster speed than serial fusion. According to the advantage above, a wavelet
decomposition feature based quaternion fusion with QEPD [1,2] for palmprint
or finger texture authentication was proposed. Also, another one with CSID [11]
migrated from face verification is implemented for palmprint or finger in this
paper. We propose a reasonable interpretation for the mathematical significance
of 4-feature parallel fusion by introducing quaternion prior to a discrimination by
such two distances respectively. Then we conduct a comparison on experimental
aspects for providing a conclusion which algorithm is better. The rest of paper
is organized as follows: section 2 introduces QEPD and CSID in turn. The next
section we propose our experimental comparison between such two distances for
handmetrics verification in details. After that, section 4 provides a conclusion.
2QEPDandCSID
2.1 Quaternion Euclidean Product Distance (QEPD)
We here make QEPD available as a discriminant distance. The relationship be-
tween QEP and quaternion modulus was discussed in [1,2]. Thus for an ar-
bitrary pixel corresponding to 4 separable wavelets decomposition sub-image,
consider two quaternions, which the former is from this pixel as the template,
P
=
a
+
bi
+
cj
+
dk
and the latter is from the tester, in which
Q
=
t
+
xi
+
yj
+
zk
. Ideally, if
P
2
, where is a particular case of Quater-
nion Euclidean product. We can estimate the difference between the template
quaternion matrix and the tester one by QEPD
D
(
PP,|PQ|
=
Q
such that
PP
=
|P |
) as a discriminant
distance. In which
is an entry of the matrix with the modulus of QEP,
and the operator
D
is certain kind of distance, e.g.
L
1
norm,
L
∞
norm distance,
Euclidean distance etc. Such two matrices have the same size as the subimage
above. The reason use the modulus is that the multiplication of two different
quaternions is a quaternion so that it is impossible to compare directly with
PP
.
Because
PQ
is not a scalar as follows.
|PQ|
PQ
=(
a − bi − cj − dk
)(
t
+
xi
+
yj
+
zk
)
=(
at
+
bx
+
cy
+
dz
)+(
ax − bt − cz
+
dy
)
i
+(
ay
+
bz − ct − dx
)
j
+(
az − by
+
cx − dt
)
k
(1)
Therefore, the similarity between template and tester QEPD
D
(
PP,|PQ|
)is
obtained by Euclidean distance between the matrix of the absolute value of the
equation (1) and that of template's modulus square
|P |
2
.
2.2 Cauchy-Schwartz Inequality Distance (CSID)
(1) Autocorrelation for signal processing.
Autocorrelation is used fre-
quently in signal processing for analyzing functions or series of values, such
as time domain signals [6]. We here discuss the case of discrete signal sequence,
defining
R
xx
(
j
)=
n
x
n
x
n−j
(2)
Search WWH ::
Custom Search