11-04-2012, 12:38 PM
Visual Cryptography for Biometric Privacy
Visual Cryptography for Biometric Privacy.pdf (Size: 1.43 MB / Downloads: 68)
INTRODUCTION
BIOMETRICS is the science of establishing the identity of
an individual based on physical or behavioral traits such
as face, fingerprints, iris, gait, and voice [1].A biometric authentication
system operates by acquiring raw biometric data from a
subject (e.g., face image), extracting a feature set from the data
(e.g., eigen-coefficients), and comparing the feature set against
the templates stored in a database in order to identify the subject
or to verify a claimed identity. The template of a person in
the database is generated during enrollment and is often stored
along with the original raw data.
SECURING IRIS AND FINGERPRINT TEMPLATES
The use of basic visual cryptography for securing fingerprint
and iris templates was suggested in [22] and [23], respectively;
however, no experimental results were reported to demonstrate
its efficacy. Moreover, basic VCS leads to the degradation in
the quality of the decoded images, which makes it unsuitable
for matching process, as shown in Fig. 10(a), where the white
background of the original image becomes gray in the decrypted
(target) image. The overlaying or superimposing operation in visual
cryptography is computationally modeled as the binary OR
operation which causes the contrast level of the target image to
be lowered.
Active Appearance Model
The proposed approach essentially selects host images that
are most likely to be compatible with the private image based
on geometry and appearance. Therefore, an active appearance
model (AAM) [25] that characterizes the shape and texture of
the face is utilized to determine the similarity between the private
face image and candidate host images (Fig. 11).
CONCLUSION AND DISCUSSION
This paper explored the possibility of using visual cryptography
for imparting privacy to biometric templates. In the
case of fingerprints and iris, the templates are decomposed
into two noise-like images using (2, 2) VCS, and since the
spatial arrangement of the pixels in these images varies from
block to block, it is impossible to recover the original template
without accessing both the shares. The XOR operator is used
to superimpose the two noisy images and fully recover the
original template.