Face recognisation system

Contents
 Face Recognition
 Steps in face recognition system
 Artificial Neural Networks as Recognizer
 Radial Basis Function Network
 Back Propagation Network
 Applications of Face Recognition System

Face Recognition
 Automatically identifying and verifying a person from an image.
 Face feature values are used for this purpose.
 Face feature values are extracted from an image and are used for
training the database using some learning method.
 Once training is complete, new images can be recognized using the
information learnt during the training process.

Steps in face recognition system:
 Face detection and tracking:
Face detection segments the face areas from the background. In case of
video, detected faces may need to be tracked using a face tracking
component.
 Face alignment:
Facial components such as eyes, nose and mouth and facial outline are
more accurately located. Based on location points , the input image is
normalized with respect to geometrical and photometrical properties.
 Feature extraction
It is performed to provide effective information that is useful for
distinguishing between faces of different persons and stable with
respect to geometrical and photometrical variations.
 Face matching
The extracted feature of input face is matched against those of
enrolled faces in database. It outputs the identity of face when a match
is found , else indicates an unknown face

Structure of Face Recognition
System

Artificial Neural Network as
Recognizer
 After extracting features from the given face image, a recognizer is
needed to recognize the face image from the stored database.
Artificial neural network can be applied for such problems.
 Two important recognition methods are:
 Radial Basis Function Network (RBF)
 Back Propagation Function Network (BPN)

Radial Basis Function Network
 The RBF network has a feed forward architecture with an input layer, a
hidden layer and an output layer.
 The input layer is simply used to take raw data and does no processing.
 The hidden layer performs a non linear mapping (using Gaussian
function)from the input space into a higher dimensional space in
which the patterns become separable.
 The output layer performs a simple weighted sum with linear output.

Back Propagation Network
 The errors propagate backwards from the output nodes to the
inner nodes.

Steps involved in back propagation network:
 Feedforward training of input patterns:
Each input node receives a signal which is broadcast to all other
hidden units. Each hidden unit computes its activation which is
broadcast to all output units.
 Back propagation of errors:
Each output node compares its activation with the desired output.
Based on the differences, the error is propagated back to all
previous inputs
 Adjustment of weights:
Weights of all links computed simultaneously based on the errors
that were propagated back.

Accuracy of RBF and BPN on the basis of training
size ratio
In case of small database BPN is more accurate but when we have large
set of images in the database that time RBF is more accurate than BPN

Applications of Face Recognition
 Preventing ID Fraud: Issuing agencies can prevent applicants from
obtaining a fraud ID by looking at photo database within seconds.
 Criminal Investigation: Face recognition technology can assist law
enforcement agencies to identify suspects.
 Physical Access Control: Authenticating authorized persons with
face recognition supports security measures in airports, stadiums,
office spaces and other buildings.

References
 A. Shekhon, P. Agarwal ; Face Recognition using Artificial Neural Network;
International Journal of Computer Science and Information Technologies,
Vol.7(2),2016.
 M. Nandini, P. Bhargavi, G. Raja Shekhar; Face Recognition using Neural
Networks, Volume3, Issue3, March 2013.
 H. A. Rowley, S Baluja , T.Kanade ; Neural Network-Based Face Detection,
Computer Vision and Pattern Recognition,1996.
 T. H. Le, Applying Artificial Neural Networks for Face Recognition, Hindawi
Publishing Corporation, Volume 2011.
 P. Latha, Dr. L. Ganeshan , Dr. S. Annadurai ; Face Recognition using Neural
Networks, Signal Processing: An International Journal, Volume 3 , Issue 5.

Face recognisation system

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Face recognisation system

Similar to Face recognisation system (20)

More from Saumya Ranjan Behura

More from Saumya Ranjan Behura (20)

Recently uploaded

Recently uploaded (20)

Face recognisation system