Information Technology Reference
In-Depth Information
same as template selection which chooses 4 templates from 8 templates in our experi-
ment. Since the comparison among the MMS, GMMS, Random and sMDIST is the
same in template selection as in OFFLINE template update procedure, we just compare
the four algorithms in experiment for ONLINE template update.
Our experiment shows that ONLINE update is more complex than OFFLINE.
Figure 13 to Figure 16 show that during ONLINE update procedure, the algorithm
GMMS is better than MMS, sMDIST and Random. Figure 17 to Figure 19 show that
GMMS can gain steady improvement than the other two algorithms.
From Figure 20 to Figure 22, we can find that OFFLINE templates update proce-
dure can gain more performance improvement than ONLINE procedure. Also, the 3
figures show that GMMS is better than MMS. From Figure 21 and Figure 22, we also
find that template update will not always improve the performance of biometric sys-
tems. It is crucial for designers to ensure the efficiency of algorithms which used in
ONLINE and OFFLINE template update procedure.
35
20
GMMS(t=1)
MMS(t=1)
sMDIST(t=1)
Random(t=1)
GMMS(t=2)
MMS(t=2)
sMDIST(t=2)
Random(t=2)
18
30
16
25
14
12
20
10
15
8
6
10
4
5
2
0
0
-2
-1
0
1
-2
-1
0
1
10
10
10
10
10
10
10
10
False Accept Rate (%)
False Accept Rate (%)
Fig. 13.
ROC curves show comparison among
MMS, GMMS, sMDIST and Random after
t=1 time update
Fig. 14.
ROC curves show comparison
among MMS, GMMS, sMDIST and Ran-
dom after t=2 times update
20
20
GMMS(t=3)
MMS(t=3)
sMDIST(t=3)
Random(t=3)
GMMS(t=4)
MMS(t=4)
sMDIST(t=4)
Random(t=4)
18
18
16
16
14
14
12
12
10
10
8
8
6
6
4
4
2
2
0
0
10
-2
10
-1
10
0
10
1
10
-2
10
-1
10
0
10
1
False Accept Rate (%)
False Accept Rate (%)
Fig. 15.
ROC curves show comparison among
MMS, GMMS, sMDIST and Random after
t=3 times update
Fig. 16.
ROC curves show comparison
among MMS, GMMS, sMDIST and Ran-
dom after t=4 times update
Search WWH ::
Custom Search