Abstract— Automatic identification of facial expressions structures the elementary nature of a variety of next generation computing devices together with sentimental computing expertise, intellectual tutoring methods, and patient sketch delicate wellness scrutinize methods etc. Therefore, we have proposed a facial expression recognition system that has the aptitude of incrementally learning and thus can learn all possible patterns of expressions that may be generated in feature. Proposed system consists of different phases including face detection, features extraction and classification. First of all, face detection has been performed by using Voila & Jones method which is robust and then transformed features has been extracted for classification using local window. Three types of features have been extracted using Discrete Cosine Transform, Haar Wavelet transform, and Gabor Wavelet. Then these features have been fused and used for classification. The results of proposed technique are compared using different quantitative measures with some of the existing techniques which show its performance.
Index Terms— Facial expression, classification.
The authors are with the Al-Imam Muhammad Ibn Saud Islamic University (IMSIU), Riyadh, Saudi Arabia (e-mail: aleisa@ccis.imamu.edu.sa, arfan.jaffar@ccis.imamu.edu.sa).
[PDF]
Cite: M. Arfan Jaffar and Eisa Al Eisa, " Classification of Facial Expression Using Transformed Features," International Journal of Information and Electronics Engineering vol. 4, no. 4, pp. 269-273, 2014.