Facial recognition using multisensor images based on localized kernel eigen spaces.

A feature selection technique along with an information fusion procedure for improving the recognition accuracy of a visual and thermal image-based facial recognition system is presented in this paper. A novel modular kernel eigenspaces approach is developed and implemented on the phase congruency feature maps extracted from the visual and thermal images individually. Smaller sub-regions from a predefined neighborhood within the phase congruency images of the training samples are merged to obtain a large set of features. These features are then projected into higher dimensional spaces using kernel methods. The proposed localized nonlinear feature selection procedure helps to overcome the bottlenecks of illumination variations, partial occlusions, expression variations and variations due to temperature changes that affect the visual and thermal face recognition techniques. AR and Equinox databases are used for experimentation and evaluation of the proposed technique. The proposed feature selection procedure has greatly improved the recognition accuracy for both the visual and thermal images when compared to conventional techniques. Also, a decision level fusion methodology is presented which along with the feature selection procedure has outperformed various other face recognition techniques in terms of recognition accuracy

---

Submitted by:
Gundimada S, Asari VK.

Symetix, Walla Walla, WA 99362, USA.

---

Abstract—A feature selection technique along with an information
fusion procedure for improving the recognition accuracy
of a visual and thermal image-based facial recognition system
is presented in this paper. A novel modular kernel eigenspaces
approach is developed and implemented on the phase congruency
feature maps extracted from the visual and thermal images individually.
Smaller sub-regions from a predefined neighborhood
within the phase congruency images of the training samples are
merged to obtain a large set of features. These features are then
projected into higher dimensional spaces using kernel methods.
The proposed localized nonlinear feature selection procedure
helps to overcome the bottlenecks of illumination variations,
partial occlusions, expression variations and variations due to
temperature changes that affect the visual and thermal face
recognition techniques. AR and Equinox databases are used for
experimentation and evaluation of the proposed technique. The
proposed feature selection procedure has greatly improved the
recognition accuracy for both the visual and thermal images when
compared to conventional techniques. Also, a decision level fusion
methodology is presented which along with the feature selection
procedure has outperformed various other face recognition techniques
in terms of recognition accuracy.
Index Terms—Feature extraction, image fusion, kernel methods,
phase congruency.
I. INTRODUCTION
MULTIBIOMETRICS refers to the use of a combination
of two or more sensor modalities in a single identification
system. The reason for combining different sensor modalities
is to improve the recognition accuracy. A multisensor biometric
system involving visual and thermal face images is presented
in this paper. Face recognition is one of the most important
applications of image analysis, its prime applications being
recognition of individuals for the purpose of security. It is one
of the most nonobtrusive biometric techniques.
Even though face recognition technology [1] has moved from
linear subspace methods [2]—Eigen and Fisher faces [3], [4]
to nonlinear methods such as kernel principal component analysis
(KPCA) and kernel Fischer discriminant analysis (KFDA)
[5]–[8], many of the problems are yet to be addressed. Also,
the nature of research studies had been more on visible imagery. However, the conclusion in [9]–[11] was
that, though the recognition performance of thermal imagery degraded,
the fusion of both visible and thermal modalities yielded
better overall performance.
Feature-based face recognition techniques [12], [13] have
demonstrated the capability of invariance to facial variations
caused by illumination and have achieved high accuracy rates.
To make the recognition process illumination invariant, phase
congruency feature maps are used instead of intensity values as
the input to the face recognition system. The feature selection
process presented in this paper is derived from the concept of
modular spaces [14]–[16]. Recognition techniques based on
local regions have achieved high accuracy rates. Though the
face images are affected due to variations such as nonuniform
illumination, expressions and partial occlusions, facial variations
are confined mostly to local regions [17]. Modularizing
the images would help to localize these variations, provided
the modules created are sufficiently small. But in this process,
a large amount of dependencies among various neighboring
pixels might be ignored. This can be countered by making the
modules larger, but this would result in an improper localization
of the facial variations. In order to deal with this problem, a
module creation strategy has been implemented in this paper
which considers additional pixel dependencies across various
sub-regions. This helps in providing additional information that
could help in improving the classification accuracy. Also, linear
subspace approaches such as PCA will not be able to capture
the relationship among more than two variables. They cannot
depict the variations caused by illuminations, expressions, etc.,
properly. In order to capture the relationships among more
than two pixels, the data is projected into nonlinear higher
dimensional spaces using the kernel method. This enables to
capture the nonlinear relationships among the pixels within the
modules.

Download full report
http://final-year-projects-chennaisundaymatlabpapers/Facial%20Recognition%20Using%20Multisensor%20Images%20Based%20on%20Localized%20Kernel%20Eigen%20Spaces.pdf

to get information about the topic Face Recognition using multisensor images based on localised kernel eigen spaces full report ppt and related topic refer the page link bellow

http://studentbank.in/report-facial-reco...gen-spaces

http://seminarsprojects.in/showthread.ph...0#pid47100

---