12-04-2012, 11:39 AM
Contactless and Pose Invariant Biometric Identification Using Hand Surface
[attachment=19965]
INTRODUCTION
HAND based biometric systems, especially hand/finger
geometry based verification systems are amongst the
highest in terms of user acceptability for biometric traits. This
is evident from their widespread commercial deployments
around the world. Despite the commercial success, several
issues remain to be addressed in order to make these systems
more user-friendly. Major problems include, inconvenience
caused by the constrained imaging set up, especially to elderly
and people suffering from limited dexterity [16], and hygienic
concerns among users due to the placement of the hand on
the imaging platform.
3-D AND 2-D HAND POSE NORMALIZATION
Fig. 1 depicts the block diagram of the proposed 3-D and
2-D hand pose normalization approach. The key idea of our approach
is to robustly fit a plane to a set of 3-D data points extracted
from the region around the center of the palm. The orientation
of the plane (normal vector) in 3-D space is then computed
and used to estimate and correct the pose of the acquired
3-D and 2-D hand.
HAND FEATURE EXTRACTION
The pose corrected range and intensity images are processed
to locate regions of interest (ROI) for hand geometry and palmprint
feature extraction. The detailed description of this method,
which is based upon the detection of interfinger points, can be
found in [15]. It may be noted that the interfinger points can be
reliably located as there can be no overlap between fingers in
the pose corrected hand images. The following section provides
a brief description of feature extraction approaches employed in
this work.
DYNAMIC FUSION
Weighted sum rule based fusion is widely employed in the
multibiometrics to combine individual match scores. The major
drawback of such a fusion framework is that poor quality samples
can have adverse influence on the consolidated score since
fixed weights are given for all samples. In order to overcome this
problem, researchers have come up with fusion approaches that
can dynamically weight a match score based upon the quality of
the corresponding modality. However, accurately computing the
quality of a biometric feature can be very challenging.
CONCLUSION
This paper has presented a promising approach to achieve
pose invariant biometric identification using hand images acquired
through a contact-free and unconstrained imaging set up.
The proposed approach utilizes the acquired 3-D hand to estimate
the orientation of the hand. The estimated 3-D orientation
information is then used to correct pose of the acquired 3-D as
well as 2-D hand. The Pose corrected intensity and range images
of the hand are further processed for extraction of multimodal
(2-D and 3-D) palmprint and hand geometry features.
We also introduced a dynamic approach to efficiently combine
these simultaneously extracted hand features.
[attachment=19965]
INTRODUCTION
HAND based biometric systems, especially hand/finger
geometry based verification systems are amongst the
highest in terms of user acceptability for biometric traits. This
is evident from their widespread commercial deployments
around the world. Despite the commercial success, several
issues remain to be addressed in order to make these systems
more user-friendly. Major problems include, inconvenience
caused by the constrained imaging set up, especially to elderly
and people suffering from limited dexterity [16], and hygienic
concerns among users due to the placement of the hand on
the imaging platform.
3-D AND 2-D HAND POSE NORMALIZATION
Fig. 1 depicts the block diagram of the proposed 3-D and
2-D hand pose normalization approach. The key idea of our approach
is to robustly fit a plane to a set of 3-D data points extracted
from the region around the center of the palm. The orientation
of the plane (normal vector) in 3-D space is then computed
and used to estimate and correct the pose of the acquired
3-D and 2-D hand.
HAND FEATURE EXTRACTION
The pose corrected range and intensity images are processed
to locate regions of interest (ROI) for hand geometry and palmprint
feature extraction. The detailed description of this method,
which is based upon the detection of interfinger points, can be
found in [15]. It may be noted that the interfinger points can be
reliably located as there can be no overlap between fingers in
the pose corrected hand images. The following section provides
a brief description of feature extraction approaches employed in
this work.
DYNAMIC FUSION
Weighted sum rule based fusion is widely employed in the
multibiometrics to combine individual match scores. The major
drawback of such a fusion framework is that poor quality samples
can have adverse influence on the consolidated score since
fixed weights are given for all samples. In order to overcome this
problem, researchers have come up with fusion approaches that
can dynamically weight a match score based upon the quality of
the corresponding modality. However, accurately computing the
quality of a biometric feature can be very challenging.
CONCLUSION
This paper has presented a promising approach to achieve
pose invariant biometric identification using hand images acquired
through a contact-free and unconstrained imaging set up.
The proposed approach utilizes the acquired 3-D hand to estimate
the orientation of the hand. The estimated 3-D orientation
information is then used to correct pose of the acquired 3-D as
well as 2-D hand. The Pose corrected intensity and range images
of the hand are further processed for extraction of multimodal
(2-D and 3-D) palmprint and hand geometry features.
We also introduced a dynamic approach to efficiently combine
these simultaneously extracted hand features.