J. Appl. Environ. Biol. Sci., 5(7)28-36, 2015
© 2015, TextRoad Publication
ISSN: 2090-4274
Journal of Applied Environmental
and Biological Sciences
www.textroad.com
*Corresponding Author: Shahid Akbar, Department of Computer Science, Abdul Wali Khan University Mardan.
shahidakbarcs@gmail.com
Face Recognition Using Hybrid Feature Space in Conjunction with Support
Vector Machine
Shahid Akbar
1
, Ashfaq Ahmad
1
, Maqsood Hayat
1
, Faheem Ali
2
1
Department of Computer Science, Abdul Wali Khan University Mardan
2
Department of Electrical Engineering University of Peshawar
Received: March 3, 2015
Accepted: May 10, 2015
ABSTRACT
Face recognition is one of the challenging problems in the area of pattern detection and recognition. It is practically
applicable in different automated systems for security purpose, access control, public security, desktop login and many
more. Due to vagueness and intricacy in face pattern, there need more exercise in order to enhance the quality of face
recognition. For this purpose, we propose a robust and reliable computational model for face recognition. In this model,
two Transformation methods such as discrete wavelet transform (DWT) and discrete sine transform (DST) along with local
based feature representation namely: local binary pattern (LBP) and local phase quantization are used to extract numerical
features from face images. Irrelevant, noisy, and redundant features are eradicated using Minimum redundancy maximum
relevance (mRMR). Various classification learners such as K-nearest neighbor (KNN), Support vector machine (SVM) and
Probabilistic Neural Network (PNN) are utilized. SUMS facial dataset and 10-folds cross validation test are used to
evaluate the performance of classification algorithms. Our proposed model achieved quite promising performance, which is
92.1% accuracy. This achievement is ascribed with the discrimination power of hybrid space and SVM. It is anticipated
that the proposed computational model might be helpful for academia and researchers in face detection and recognition.
KEYWORDS: DST, DWT, LBP, LPQ, KNN, SVM, PNN.
1. INTRODUCTION
In order to develop an efficient and reliable recognition system a lots of efforts have been carried out by the
researchers for face detection [1; 2; 3] and individual identification [4]. Face recognition is one of the few biometric
methods that possess the merits of both high accuracy and low intrusiveness. It’s used in many areas such as entertainment,
information security and surveillance [5].Hoang Le et al., used two dimensional principal component analysis approach to
extract features from both FERET and AT&T facial datasets. The recognition rate of the proposed system is evaluated
through SVM and KNN [6]. Chengjun et al., proposed Gabor–Fisher classifier (GFC) for face recognition. Augmented
Gabor based features vector is derived through the Gabor wavelet representation of the face images. In order to get high
discriminative features, the extracted features space is reduced using Enhanced Fisher linear Discriminant Model
(EFM).The comparative analysis of the GFC with all the considerable techniques reveals that proposed GFC achieved the
high recognition results using FERAT frontal faces images [7].similarly, Yilmaz et al., applied a novel preprocessing
technique “Eigen Hills”. In this approach Eigen face based features is extracted from facial images, but the performance
results did not achieved the expected results due to expression variation [8; 9]. Furthermore, Jagadeesh et al., has proposed
“DBC-FR” algorithm for face recognition. Proposed technique is tested on NIR face image database. In order to match the
extracted features, Euclidian distance is utilized for training and testing purpose. It is observed that proposed method
obtained better success rates than existing techniques [10]. Likely, Nayak et al., used fused discrete cosine transform and
discrete wavelet transform to extract feature spaces from ORL database. The performance of the proposed system is
measured using SVM and ANN [11]. Wadkar et al., applied haar wavelet transform and Biorthogonal wavelet transform
based techniques on ORL facial dataset. The comparative analysis of the proposed shows that haar wavelet transform
obtained better results than that of biorthogonal wavelet transform [12]. Hengliang et al., used HOG filter on the
normalized images. Both local and global HOG features are extracted by combining Principal component analysis and
linear discriminant analysis [13]. Moreover, local based feature based approaches received the attention of the researchers
for facial detection from last few years [14]. LBP is a non-parametric technique that assembles efficient information
about local structure of face image and recognizes an individual based on gather features [15; 16]. Ahonen et al., divided
the face image into several region. High discriminative descriptors are extracted from each sub parts of the image and
concatenated into a single feature vector. The recognition rate is evaluated using nearest neighbor classifier. The proposed
scheme clearly shows the superiority over all the compared methods [17].
28