dc.description.abstract | Recently, personal verification based on biometric features
gradually becomes an important and highly demand technique for
security access systems. During the past, numerous literatures discussing
biometric verification using palm features have been reported. However,
they are all constrained by some limitations, such as the utilization of
docking devices to constrain the palm position while acquiring palmprint
images, the applying of inked palmprint images as the objects, and the
requirements of adequate lighting conditions, etc. These limitations
hinder the conveniences of users and the practicalities of verification
methods. In this dissertation, novel methods are devised and developed
to alleviate or remove theses limitations. In our work, the limitation
imposed by the docking devices is removed completely. Furthermore,
the inconveniences introduced by the applying of inked palmprint
images as objects are avoided. Finally, the restrictions of lighting
conditions are also avoided.
In this dissertation, we propose two approaches of biometric
verification based on the palm features, which are principal palmprints
and vein-patterns of palm-dorsa, respectively. The crucial characteristics
of the proposed methods are that no prior knowledge about the objects is
necessary, the parameters can be set automatically, and the limitations as
mentioned above can be alleviated.
In the palmprint verification approach, scanner is adopted as the
input device for capturing palmprint images with the advantages of no
palm inking and no requirement of docking device. Two finger-webs are
automatically selected as the datum points to define the Region of
Interest (ROI) in the palmprint images. Next, hierarchical decomposition
is employed to extract the principal palmprint features inside the ROI,
which includes direction and multiresolution decompositions. The
former extracts principal palmprint features from each ROI. The latter
processes the images with principal palmprint features to extract the
dominant points from the images at each resolution. Finally, normalized
correlation function is utilized to evaluate the similarity between two
palmprint images. Experiments were conducted on a wide variety of
palmprint images and the results are satisfactory with acceptable
accuracy rate (FRR: 0.75% and FAR: 0.56%). The results reveal that our
proposed approach is feasible and effective in palmprint verification
without the needs of docking devices or palm inking.
In the vein-patterns verification approach, an infrared (IR) camera
is adopted as the input device to capture the thermal images of
palm-dorsa. Likewise, two of the finger-webs are automatically selected
as the datum points to define the Region of Interest (ROI) on the thermal
images. Within each ROI, feature points of the vein-patterns (FPVPs) are
extracted by modifying the basic tool of watershed transformation based
on the properties of thermal images. According to the heat conduction
law (the Fourier law), multiple features can be extracted from each
FPVP for verification. Multiresolution representations of images with
FPVPs are obtained using multiple multiresolution filters (MRFs) that
extract the dominant points by filtering miscellaneous features for each
FPVP. A hierarchical integrating function is then applied to integrate
multiple features and multiresolution representations. The former is
integrated by an inter-to-intra personal variation ratio and the latter is
integrated by a stack filter. We also introduce a logical and reasonable
method to select a trained threshold for verification. The proposed
approach can achieve an acceptable accuracy rate (FRR: 2.3% and FAR:
2.3%). The experimental results demonstrate that our proposed approach
is valid and effective for vein-pattern verification. | en_US |