Latest International Finger/Palm Print Test Result of Hisign-- Updated on 20170728
1 NIST Test
1.1 MINEXIII Test
The MINEX III evaluation is a NIST SP 800-76-2 Biometric Specifications for Personal Identity Verification (PIV) U.S. Government program compliance evaluation. It is a successor to the discontinued MINEX Ongoing program.
We have passed the Generator test of MINEXIII, the template generated conformed to NIST requirements. According to the result, Hisign ranks the fourth.
We have passed the Matcher test of MINEXIII and ranked at the sixth.
Related web
link:
Template
Generator Report
http://nigos.nist.gov:8080/evaluations/minexiii/reportcards/005B+0015_generator_report.pdf
Template
Matcher Report
http://nigos.nist.gov:8080/evaluations/minexiii/reportcards/005B+0015_matcher_report.pdf
Top Vendors
Results
1.2 PFTII Test
NIST Proprietary Fingerprint Template Evaluation II (PFT II)
is one-to-one verification evaluation which measures the performance of
fingerprint matching algorithms by utilizing proprietary fingerprint templates.
The samples dataset have been increased to 120,000 subjects compared to
previous PFT evaluation. Number of experiments was also increased to 33 with
different combinations of single and two fingerprints matching.
In 2017 Hisign
fingerprint algorithm was submitted to NIST Proprietary Fingerprint Template
Evaluation II. The algorithm's template matching accuracy was among the best
participants in most of the experiments.
The test
result for each vendors can be found at https://www.nist.gov/itl/iad/image-group/proprietary-fingerprint-template-evaluation-ii-pftii-results, and the data
in below sheets are the summary of combined and comparison of each vendors test
results.
1.2.1 Template Size
There is no
limitation to the size of templates in the test. Therefore, template size
varies significantly among the participant vendors. Theoretically, the increase
in template size can pose positive effect on the accuracy, while matching time
required is also extended. The algorithm submitted by Hisign is based on
practice so that the template size, and hence the registration time and
matching time are balanced for overall performance, especially for super large
fingerprint database. Below sheet shows the average template sizes of Hisign in
different databases.
Database
Hisign
Median Template Size(Byte)
AZGallery
7254
AZProbe
7186
LAGallery
4612
LAProbe
5142
DHS2Gallery
3220
DHS2Probe
3216
POEBVAGallery
3208
POEBVAProbe
3506
Click here for
the average template sizes of various vendors in different databases
1.2.2 EER
Performance of
Hisign algorithm is listed in below sheet.
Type
Database
Pose
Hisign
Rank
P2P
POEBVA
02
0.00177
2/16
P2P
POEBVA
07
0.00473
6/16
P2P
POEBVA
02+07
0.00020
2/16
P2P
DHS2
02
0.01307
4/15
P2P
DHS2
07
0.01228
4/15
P2P
DHS2
02+07
0.00687
5/16
P2P
AZLA
01
0.00188
4/15
P2P
AZLA
02
0.00385
3/15
P2P
AZLA
03
0.00526
2/15
P2P
AZLA
06
0.00194
6/15
P2P
AZLA
07
0.00526
2/15
P2P
AZLA
08
0.00696
3/15
P2P
AZLA
01+06
0.00066
5/16
P2P
AZLA
02+07
0.00071
2/16
P2P
AZLA
03+08
0.00144
2/16
P2R
AZLA
01
0.00196
6/14
P2R
AZLA
02
0.00322
3/15
P2R
AZLA
03
0.00396
3/15
P2R
AZLA
06
0.00224
6/14
P2R
AZLA
07
0.00441
3/15
P2R
AZLA
08
0.00531
4/15
P2R
AZLA
01+06
0.00074
5/15
P2R
AZLA
02+07
0.00077
3/16
P2R
AZLA
03+08
0.00112
2/16
R2R
AZLA
01
0.00153
7/14
R2R
AZLA
02
0.00141
4/14
R2R
AZLA
03
0.00165
3/14
R2R
AZLA
06
0.00158
6/14
R2R
AZLA
07
0.00184
3/14
R2R
AZLA
08
0.00195
2/14
R2R
AZLA
01+06
0.00051
7/15
R2R
AZLA
02+07
0.00049
2/15
R2R
AZLA
03+08
0.00056
5/15
Click here for
the comparison of EER of the participant vendors
Note: NA means
data are not obtained, and similarly hereinafter.
1.2.3 FNMR@FMR=0.0001
Performance of
Hisign algorithm is listed in below sheet.
Type
Database
Pose
Hisign
Rank
P2P
POEBVA
02
0.00266
3/17
P2P
POEBVA
07
0.00781
4/17
P2P
POEBVA
02+07
0.00022
2/17
P2P
DHS2
02
0.002191
4/17
P2P
DHS2
07
0.02074
4/17
P2P
DHS2
02+07
0.01353
3/17
P2P
AZLA
01
0.00292
7/17
P2P
AZLA
02
0.00824
3/17
P2P
AZLA
03
0.00904
3/17
P2P
AZLA
06
0.00275
6/17
P2P
AZLA
07
0.00963
2/17
P2P
AZLA
08
0.01068
2/17
P2P
AZLA
01+06
0.00075
7/17
P2P
AZLA
02+07
0.00103
3/17
P2P
AZLA
03+08
0.00198
2/17
P2R
AZLA
01
0.00297
7/17
P2R
AZLA
02
0.00621
3/17
P2R
AZLA
03
0.00746
3/17
P2R
AZLA
06
0.00340
6/17
P2R
AZLA
07
0.00840
3/17
P2R
AZLA
08
0.00827
3/17
P2R
AZLA
01+06
0.00077
4/17
P2R
AZLA
02+07
0.00100
3/17
P2R
AZLA
03+08
0.00153
3/17
R2R
AZLA
01
0.00193
5/17
R2R
AZLA
02
0.00178
5/17
R2R
AZLA
03
0.00209
3/17
R2R
AZLA
06
0.00201
7/17
R2R
AZLA
07
0.00228
2/17
R2R
AZLA
08
0.00239
2/17
R2R
AZLA
01+06
0.00052
6/17
R2R
AZLA
02+07
0.00051
5/17
R2R
AZLA
03+08
0.00057
2/17
Click here for the comparison of FNMR@FMR=0.0001
performance of the participant vendors
1.2.4 FMR@FNMR=0.002
Performance of
Hisign algorithm is listed in below sheet.
Type
Database
Pose
Hisign
Rank
P2P
POEBVA
02
0.00071
2/14
P2P
POEBVA
07
0.24074
6/14
P2P
POEBVA
02+07
NA
NA
P2P
DHS2
02
0.44624
2/10
P2P
DHS2
07
0.40382
3/10
P2P
DHS2
02+07
0.04455
4/12
P2P
AZLA
01
0.00131
4/13
P2P
AZLA
02
0.02587
3/11
P2P
AZLA
03
0.22884
4/10
P2P
AZLA
06
0.00153
5/13
P2P
AZLA
07
0.11432
4/11
P2P
AZLA
08
0.48283
4/10
P2P
AZLA
01+06
NA
NA
P2P
AZLA
02+07
NA
NA
P2P
AZLA
03+08
0.00008
1/14
P2R
AZLA
01
0.00181
4/13
P2R
AZLA
02
0.01376
3/11
P2R
AZLA
03
0.04025
3/11
P2R
AZLA
06
0.00578
7/13
P2R
AZLA
07
0.07824
3/11
P2R
AZLA
08
0.29759
5/10
P2R
AZLA
01+06
NA
NA
P2R
AZLA
02+07
NA
NA
P2R
AZLA
03+08
NA
NA
R2R
AZLA
01
0.00007
2/11
R2R
AZLA
02
NA
NA
R2R
AZLA
03
0.00022
4/13
R2R
AZLA
06
0.00013
3/12
R2R
AZLA
07
0.00055
3/13
R2R
AZLA
08
0.00164
3/13
R2R
AZLA
01+06
NA
NA
R2R
AZLA
02+07
NA
NA
R2R
AZLA
03+08
NA
NA
Click here for the comparison of FNMR@FMR=0.0002
performance of the participant vendors
1.2.5 FMR@FNMR=0.005
Performance of
Hisign algorithm is listed in below sheet.
Type
Database
Pose
Hisign
Rank
P2P
POEBVA
02
NA
NA
P2P
POEBVA
07
0.00318
3/14
P2P
POEBVA
02+07
NA
NA
P2P
DHS2
02
0.16714
3/12
P2P
DHS2
07
0.14718
3/12
P2P
DHS2
02+07
0.01591
4/12
P2P
AZLA
01
NA
NA
P2P
AZLA
02
0.00120
3/13
P2P
AZLA
03
0.00703
3/14
P2P
AZLA
06
NA
NA
P2P
AZLA
07
0.00650
3/13
P2P
AZLA
08
0.03088
4/14
P2P
AZLA
01+06
NA
NA
P2P
AZLA
02+07
NA
NA
P2P
AZLA
03+08
NA
NA
P2R
AZLA
01
NA
NA
P2R
AZLA
02
0.00038
3/12
P2R
AZLA
03
0.00125
2/13
P2R
AZLA
06
NA
NA
P2R
AZLA
07
0.00250
3/14
P2R
AZLA
08
0.00752
4/13
P2R
AZLA
01+06
NA
NA
P2R
AZLA
02+07
NA
NA
P2R
AZLA
03+08
NA
NA
R2R
AZLA
01
NA
NA
R2R
AZLA
02
NA
NA
R2R
AZLA
03
NA
NA
R2R
AZLA
06
NA
NA
R2R
AZLA
07
NA
NA
R2R
AZLA
08
NA
NA
R2R
AZLA
01+06
NA
NA
R2R
AZLA
02+07
NA
NA
R2R
AZLA
03+08
NA
NA
Click here for the comparison of FNMR@FMR=0.0005
performance of the participant vendors
1.2.6
Summary
Based on above
data, Hisign achieved superior performance in most of the tests that the
comprehensive ranking is within top level among the participant vendors.
2
FVC-onGoing Test
The aim is to
track the advances in fingerprint recognition technologies, through
continuously updated independent testing and reporting of performances on given
benchmarks. The algorithms are evaluated using strongly supervised approaches
to maximize trustworthiness.
2.1
ISOMatch
2.2 FvMatch
FvMatch is a
test of the performance of the participants’ matching algorithm on their own
characteristics.
STD mode and
HARD mode: ranked the first at the time of this news release.
(Ranking may be changed after updated date.)
2.3 Palm
Verification
Palm
verification is a test of the performance of the participants’ matching
algorithm on their own characteristics.
All
benchmarks: ranked the second at the time of this news release. (Ranking may be
changed after updated date.)
Related Web
link:
Click Here for the
published results. 1. Click on one of the three test thumbnails to the left,
and then; 2. Select benchmark type; 3. Click on the
EER/FMR1000/FMR10000 column head to view corresponding rankings.
FVC-onGoing is
a web-based automated evaluation system for fingerprint recognition algorithms.
Tests are carried out on a set of sequestered datasets and results are reported
on-line by using well known performance indicators and metrics.
ISOMatch
contains fingerprint matching benchmarks using a standard minutiae-based
template format [ISO/IEC 19794-2 (2005)]. The test uses two data sets, a
standard test set and a low quality (hard) test set. The standard test set
images are of good quality fingerprint images. The quality of the low quality
test is low, and there are more noise and deformation. Hisign's algorithm
ranked second for low quality test (HARD mode) and seventh for standard test
(STD mode) at time of this news release. (Ranking may be changed
after updated date.)
京公网安备 11010602007531号
CopyRight © 北京海鑫科金高科技股份有限公司