1.
Multi-biometric Cryptosystems
Based on Feature-Level Fusion
Presented By
2.
Outline
Literature Survey
Objective
Significance
Proposed Methodology
S/W and H/W Technology
Mathematical Model
3.
Literature survey
Sutcu combined face and fingerprint templates that are both transformed into
binary strings. These binary strings are concatenated and used as the input
to a fuzzy commitment scheme.
Nanda-kumar and Jain proposed a multi-biometric cryptosystem in which
biometric templates based on binary strings and point-sets are combined.
The binary string is divided into a number of segments and each segment is
separately secured using a fuzzy commitment scheme. The keys associated
with these segment-wise fuzzy commitment schemes are then used as
additional points in the fuzzy vault constructed using the point-set-based
features.
Fu theoretically analyzed the template security and recognition accuracy
imparted by a multi-biometric cryptosystem, which can be operated in four
different ways: no-split, MN-split, package, and biometric model.
4.
Objective
To perform and analyze research work on Multi-biometric cryptosystems
that uses the knowledge from the areas of statistical pattern
recognition, biometric recognition, computer vision and image processing and
develop innovative solutions that allow secure, convenient, and. cost-
effective interaction between humans and machines.
5.
Significance
Multi-biometrics are used world-wide because there will be low error rate and
large population coverage.
The multibiometric cryptosystems proposed here have higher security and
matching performance compared to their unibiometric counterparts.
Multibiometric systems are being widely adopted inmany large-scale
identification systems, including the FBI’s IAFIS, the Department of
Homeland Security’s US-VISIT, and the Government of India’s UID.
6.
Proposed Methodology
In our study we propose a feature level framework for multi-biometric multi-
systems that consists of the three basic modules :
Embedding Algorithm
Fusion Module
Biometric Multi-systems
Security Analysis
7.
Embedding Algorithm
In embedding algorithm, The embedding algorithm
transforms a biometric feature representation into a
new feature representation zm , where zm= Σ (m)xm
, for all m=1,2,……M. The input representation can
be a real-valued feature vector, a binary string, or a
point-set. The output representation could be a
binary string or a point-set that could be secured
using fuzzy commitment or fuzzy
vault, respectively.
8.
Fusion Module
The fusion module combines a set of homogeneous
biometric features Z={z1,z2,z3…zm} to generate a fused
multibiometric feature representation z. For point-set-based
representations, one can use Cs(Z)=Um=1 Zm. In the case of
binary strings, the fused feature vector can be obtained by
simply concatenating the individual
strings, i.e., z=Cb(Z)=[Z1,Z2,Z3…Zm]. Note that it is also
possible to define more complex fusion schemes, where
features could be selected based on criteria such as reliability
and discriminability.
9.
Biometric Cryptosystems
In biometric cryptosystem , the biometric
cryptosystem generates a secure sketch Yc the
fused feature vector Ze(obtained from the set of
biometric templates ) and a key , i.e. During
authentication, the biometric templates XE={ X1E
,X2E ……XE
M} and a key kc .During cryptosystem
recovers Kc from Yc and ZA. Fuzzy commitment is
used if is a binary string, whereas a fuzzy vault is
used if is a point-set.
10.
Security Analysis
Each of the above three modules play a critical role
in determining the matching performance and
security of the multi-biometric cryptosystem. We
propose fairly simple algorithms for implementing
the above three modules and do not focus on
optimizing them.
11.
H/W and S/W requirement
Hard disk 80 GB
RAM 512 MB
Processor Intel Pentium IV
Technology Java
Tools Net-Beans IDE + JDK
Operating System Windows
Seja o primeiro a comentar