An Improved Face Detection and Recognition Security System

advertisement
International Journal of Engineering Trends and Technology (IJETT) – Volume 4 Issue 9- Sep 2013
An Improved Face Detection and Recognition
Method Based On Information of Skin Colour for
Security System
Srivalli P#1, B Karunaiah*2, K V Murali Mohan*3
1
2
Srivalli P, pursuing M.Tech (ECE) from Holy Mary Institute of Technology and science (HITS), Bogaram, Keesara,
Hyderabad. Affiliated to JNTUH, Hyderabad, A.P, INDIA
B Karunaiah, working as a Professor (ECE) at Holy Mary Institute of Technology and science (HITS), Bogaram, Keesara,
Hyderabad. Affiliated to JNTUH, Hyderabad, A.P, INDIA
3
K V Murali Mohan, working as a Professor and HOD (ECE) at Holy Mary Institute of Technology and science (HITS),
Bogaram, Keesara, Hyderabad. Affiliated to JNTUH, Hyderabad, A.P, INDIA
Abstract— In lot of areas, unauthorised entry is prohibited to
have an assurance that something valuable is not stolen. This
paper employs two of the emerging technologies- Digital Image
Processing and Embedded Systems to develop a prototype so that
the door can be unlocked only by recognising the faces of
authenticated persons. The electronic Embedded Systems which
consists of a camera, is interfaced with the PC for capturing and
processing images.
Keywords— Eigen faces, Templates, Face Segmentation, Gabor
filters.
I.
INTRODUCTION
Security is an important feature in all environments.
Security relates to protection of anything from unwanted
people. Safety is one of the elements that complete one’s
requirement for a meaningful life that leads to selfactualization [1, 2, 3, and 4]. Deeming all the many ill fated
cases connected to security lapses in organizations, the
effortlessness and benefits of computer based security devices
has brought the use of computer controlled security door
systems in the modern vocabulary. Recent research works
have shown that diverse human characteristics like finger
prints and voice tags have been used for the development of
an automated security system but further studies have revealed
that a Sophisticated Security System can be developed using
facial recognition systems.
Face recognition is a famous image processing techniques
for bio-metric based security system and humans frequently
use faces to recognize individuals and improvement in
ISSN: 2231-5381
computing capability over the last few decades now enable
similar recognitions automatically. Earlier the face recognition
algorithms used plain geometric models, but the recognition
process has now changed into a science of complicated
mathematical representations and matching procedures. The
new research and proposals in the past ten to fifteen years
have driven face recognition technology into the focus. Face
recognition can be used for both verification and identification.
Above all computational models of face recognition are
appealing because they can contribute not only to theoretical
insights but also to practical applications. Unfortunately,
developing a computational model of face recognition is quite
difficult, because faces are intricate, multidimensional and
motivation for important visual factors. They are a natural
class of objects, and stand in severe contrast to sine wave
gratings, the “blocks world,” and other artificial stimuli used
in human and computer vision research [5]. Thus unlike the
majority of early visual functions, for which we may build-up
a full representation of retinal or striate behaviour, face
recognition is a high level task for which computational
approaches can at present only suggest broad limitations on
the corresponding neural activity.
In some research works, an associative network with a
simple learning that could classify face images and recall a
face image from an incomplete or noisy version input to the
network was described [6 and 7]. Our approach is based on an
information theory approach that decomposes face images
into a small set of characteristic feature images called
http://www.ijettjournal.org
Page 3851
International Journal of Engineering Trends and Technology (IJETT) – Volume 4 Issue 9- Sep 2013
‘Eigenfaces’ which are actually the principal components of
the initial training set of facial images. Recognition is
accomplished by projecting a new image into the subspace
spanned by the Eigenfaces (‘face space’) and then classifying
the face by comparing its position in the face space with the
positions of the known individuals. Eigenfaces are the Eigen
vectors which are representative of each of the dimensions of
this face space and they can be considered as facial features. A
face can be represented as linear function of the singular
vectors of the set of faces, and then singular vectors are eigen
vectors of the covariance matrices.
Generally, face recognition techniques is divided into two
groups based on the face representation they use:
1) Appearance based uses holistic texture features
and is applied to either whole face or specific
regions in a face image.
2)
Feature based uses geometric facial features
(mouth, eyes, cheeks etc.) and geometric relations
between them.
Among many approaches to the problem of face
recognition have been proposed, subspace analysis based on
appearance is one among the oldest, and even so it yields the
most encouraging results. Subspace analysis is executed by
projecting an image into a lower dimensional space and after
that recognition is performed by measuring the distance
between known images and the image to be identified. The
most difficult part of such a system is finding a subspace.
II.
EIGENFACE APPROACH
A. Eigen Values and Eigen Vectors
In linear algebra, the eigen vectors of a linear operator are
non-zero vectors which, when operated on by the operator,
give a scalar multiple of them. The scalar is then called the
Eigen value (λ) associated with the Eigen vector (X). Eigen
vector is a vector that is scaled by a linear transformation.
This is an inherent property of a matrix. Whenever a matrix
operates on it, only the magnitude of the vector is changed but
the direction remains the same
=
Where A is a vector function.
Calculation of Eigen Values and Eigen Vectors
By using the above equation
ISSN: 2231-5381
( −
) =0
Where I is the n x n identity matrix. From fundamental
linear algebra it is clear that a nontrivial solution exists only
and only if
( −
)=0
Where det denotes determinant. When evaluated, it
becomes a polynomial of degree n. This is known as the
characteristics equation of the matrix A and the corresponding
polynomial is known as characteristics polynomial. Thus there
are n eigen values of A satisfying the equation
AXi=λXi
Where i=1, 2, 3...n.
In the case where there are r repeated eigenvalues, then a
linearly independent set of n eigen vectors exist, provided the
rank of the matrix (A-λI) is rank n-r.
Once the eigen values of the matrix has been found, eigen
vectors can be found by Gaussian elimination.
III.
PCA
A. .Overview of the PCA algorithm:
The algorithm for the facial recognition using eigenfaces is
basically described First, the original images of the training
set are transformed into a set of eigenfaces E. Later; the
weights are calculated for each image of the training set and
stored in a set W. After observing an unknown image X, the
weights are evaluated for that specific image and stored in the
vector WX. Later, WX is compared with the weights of
images, of which we certainly have knowledge that they are
facing (the weights of the training set W). One method of
achieving this would be to regard each weight vector as a
point in space and calculate an average distance between the
weight vectors from WX and the weight vector of the e image
WX. If this average distance exceeds some threshold value ,
then the weight vector of the test image WX lies too ”far apart”
from the weights of the faces. In this case, the test image X is
considered to not a face. Otherwise (if X is actually a face), its
weight vector WX is stored for further categorization. The
optimal threshold value has to be calculated empirically. By
using PCA we can transform each original image of the
training set into a corresponding Eigen face. A significant
property of PCA is that we can reconstruct any original image
from the training set by combining the Eigen faces. Talking of
Eigenfaces, we can say that they are nothing less than
http://www.ijettjournal.org
Page 3852
International Journal of Engineering Trends and Technology (IJETT) – Volume 4 Issue 9- Sep 2013
characteristic attributes of the faces. So we can say that the
original face image can be reconstructed from Eigen faces.
IV.
Significance of Eigenface Approach:
We have seen the standard Eigenface Approach for
Face Recognition. In this work, we find out M Eigenvectors
corresponding to training set images. Now choosing only M'
Eigenvectors from these M Eigenvectors is the most important
task. This selection is done in a way such that M' is less than
M, to symbolize face space spanned by images. This will
diminish the face space dimensionality and increase speed of
face recognition. We can choose only M' Eigenvectors with
maximum Eigen values. Now as the higher Eigen values
represent maximum face variation in the corresponding
Eigenvector direction, considering this Eigenvector for face
space representation is important. As the lower Eigen values
do not provide much information about face variations in
corresponding Eigenvector direction, these small Eigen values
are neglected to further decrease the dimension of face space.
This does not reduce the success rate much and is acceptable
depending on the application of face recognition. Here only
M' Eigenvectors with higher Eigen values are chosen for
defining face space. The most critical task in this process is to
choose value of M' (Maximum Eigen values chosen from Covariance Matrix) so that it does not result in high error rates
for Face identification process. The M' must be selected such
that error rates do not increase much and are acceptable
depending on application for face recognition.
V.
KIT PROTOTYPE
VI.
CONCLUSION
The Eigenface approach for face recognition process is fast
and simple and works well under constrained environment. It
gives speedy results which can be immediately deciphered by
the Embedded prototype kit and the door operation can be
controlled. For given set of images, due to high
dimensionality of images, the space spanned is enormous. But
as a matter of fact, all these images are directly related and
actually span a lower dimensional storage. Using Eigen face
approach reduces this dimensionality. The lower the
dimensionality of the image space, the easier it is for face
recognition. Any new face can be expressed as a linear
combination of these eigen faces. This makes it easier to
match any two faces and thus complete face recognition.
We have developed a prototype of a door locking system
where an electromagnetic relay is being used to represent the
door locking system. This relay can be later connected to a
solenoid driver to arrange a real time system. An Embedded
PIC micro controller was used to collect the face recognition
results from the MATLAB software through RS232
communication. An Embedded C program was developed and
burned in the uC flash memory so that whenever the face
recognition results are found to be successful, the uC will be
able to drive the relay in order to open the door. After a
certain delay, the relay is again brought back to open
connection by the micro controller in order to close the door.
ISSN: 2231-5381
http://www.ijettjournal.org
Page 3853
International Journal of Engineering Trends and Technology (IJETT) – Volume 4 Issue 9- Sep 2013
VII.
[1]
REFERENCES
Ahmad Yusairi Bani Hashim, Zamberi Jamaludin, “An Enduring
Relationship between Biometric Traits and Security System”,
International Journal of Research and Reviews in Information’s
Security and Privacy (IJRRISP) Vol. I, No. 1, March 2011.
[2]
A.H. Maslow, The Farther Needs of Human Nature. New York:
Penguin Group, 1971.
[3]
A.H. Maslow, Towards the Psychology of Being Second Edition New
York: Van Nostrand Reinhold, 1968.
[4]
A.H. Maslow, Religion, Values and Peak Experiences, New York:
Penguin Group, 1970.
[5]
Davies, Ellis, & Shepherd (EDS.), “Perceiving and Remembering
Faces”, London: Academic Press, 1981.
[6]
T. Kohonen, “Self organization and associative memory”, Berlin:
Springer-Verlag (1989).
[7]
T. Kohonen, P. Lehtio, “Storage and Processing of Information in
Distributed Associative Memory Systems”, in G.E. Hinton & J.A.
Anderson (Eds.), “Parallel Models of Associative Memory,” Hillsdale,
New Jersey: Erlbaum, pp. 105-143. 1981.
[8]
Tae-Kyum Kim, Josef Kittler, Roberto Cipolla “Discriminative
learning and Recognition of image set classes using Canonical
correlation,” Published in IEEE Trans. On Pattern Analysis and
Machine Intelligence, June 2007, pp 1005-1018
[9]
S.T. Gandhe, K T Talele, A G Keskar, “Intelligent Face Recognition: A
Comparative Study,” published in GVIP journal, Volume 7, Issue 2,
August, 2007.
[10]
N Krishnan, GR Jothi, K Pulloor, GL Aaron, “Recognition of NonSymmetric faces using Principal Component Analysis,” ICVGIP
Proceedings, October 2012.
[11]
M.A. Turk, A.P. Pentland, “Face Recognition using Eigenfaces,”
Published in IEEE Computer Society Conference on Computer Vision
and Pattern Recognition, 3-6 June 1991, pp 586-591.
[12]
Matthew Turk, Alex Pentland, “Eigenfaces for Recognition,” Vision
and Modeling Group, The Media Laboratory, Massachusetts Institute
of Technology, in J. of cognitive neuroscience, 1991, vol. 3, no. 1,
pages 71 to 86.
[13]
A.M. Martinez and A.C. Kak, “PCA versus LDA,” IEEE Trans.
Pattern Analysis and Machine Intelligence, vol. 23, no. 2, pp. 228–233,
Feb. 2001 Nick Pears Thomas Heseltine and Jim Austin. “ Evaluation
of image preprocessing techniques for eigenface based face
recognition”. ACA Group, Dept. of Computer Science, University of
York, 2002.
ISSN: 2231-5381
http://www.ijettjournal.org
VIII.
AUTHOR DETAILS
Srivalli P, Pursuing
MTech (ECE) from Holy
Mary
Institute
of
Technology and science
(HITS),
Bogaram,
Keesara,
Hyderabad.
Affiliated to JNTUH,
Hyderabad, A.P, INDIA
B Karunaiah, working
as a Professor (ECE) at
Holy Mary Institute of
Technology and science
(HITS),
Bogaram,
Keesara,
Hyderabad.
Affiliated to JNTUH,
Hyderabad, A.P, INDIA
K V Murali Mohan,
working as a Professor
and HOD (ECE) at Holy
Mary
Institute
of
Technology and science
(HITS),
Bogaram,
Keesara,
Hyderabad.
Affiliated to JNTUH,
Hyderabad, A.P, INDIA
Page 3854
Download