A UNIFIED APPROACH TO EMOTION RECOGNITION USING HAAR LIKE AND GABOR WAVELET

Abstract

Human emotional behavior recognition has become the key topic in the research domain. A large number of techniques and tactics are applied in order to obtain the high accurate mechanism for emotion detection. The mechanisms like classification, feature extraction, fusion and others are the major tactics that can be utilized in order to enhance the accuracy and reduce the error rate of the emotion recognition system. This study is organized with an objective to develop an emotion recognition system by using Gabor and Haar-like feature extraction algorithm, PCA for image fusion, SVM, ANN (Feed Forward Neural network) and KNN classifiers for classification analysis. For the purpose of simulation, the MATLAB simulation platform is utilized and testing of the proposed work is done by using three different datasets i.e. CohnKanade, Yale and Jaffe dataset. The proficiency of proposed emotion recognition technique is evaluated in the terms of Accuracy Rate. After evaluating the results, the study proves that the ANN classifier has the highest accuracy rate in comparison to the KNN and SVM classifiers.