A Hybrid Feature Extraction Approach for Face Recognition Systems

 

A. Saradha (1) and S. Annadurai (2)
(1) Senior Lecturer in CSE, Institute of Road and Transport Technology, Erode, Tamilnadu, India.
(2) Principal, Government College of Engineering, Tirunelveli, Tamilnadu, India

Abstract:

Automatic recognition of individuals is a significant problem in the field of pattern recognition. The face images considered for recognition undergo large variations due to changes in illumination conditions, viewing direction, facial expression and aging etc. The face images also have similar geometrical features and hence discriminating one face from the other in the database is a challenging task. Hence it is very difficult to represent face images with distinct feature vectors that are invariant to transformation. Even though the extracted feature vectors possess overlapping characteristics, the problem may be easily solved if there exists a feature extraction method which can generate distinct features for each class of image or a classification technique capable of discriminating the overlapping features of the images. In this paper feature extraction techniques such as Fourier descriptors, Zernike moments, Hu moments and Legendre moments are considered and classification techniques such as Nearest Neighbor classifiers, Linear Discriminant Analysis classifiers and neural network classifiers are compared. From the comparative study the most suitable feature extraction approach and classification algorithms are identified for face recognition. All the feature extraction methods are tested with ORL database of 40 subjects and each of them with 10 orientations. The performance comparison is made among different approaches.

 

 

BibTex: