Facial Features Based Hybrid Methods for Emotion Recognition
Bharati Dixit1, Arun Gaikwad2
1Bharati Dixit, Associate Professor, Department of Computer Engineering, MIT College of Engineering, Pune and Research Scholar, SCOE, Pune (Maharashtra), India.
2Dr. Arun Gaikwad, Professor and Former Principal, Department of Electronics and Telecommunications, ZCER, Pune (Maharashtra), India.
Manuscript received on 19 September 2019 | Revised Manuscript received on 06 October 2019 | Manuscript Published on 11 October 2019 | PP: 385-389 | Volume-8 Issue-2S10 September 2019 | Retrieval Number: B10650982S1019/2019©BEIESP | DOI: 10.35940/ijrte.B1065.0982S1019
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Effective human machine interaction systems are need of the time so the work carried out deals with one of such significant HMI tasks- automatic emotion recognition. The experimentation carried out for this study is focused to facial expressions based emotion recognition. Two techniques of emotion recognition based on hybrid features are designed and experimented using JAFFE database. The first technique referred as “Hybrid Method1” is designed around feature descriptor obtained through local directional number & principal component analysis and feed forward neural network used as classifier. The second technique referred as “Hybrid Method 2” is designed around feature descriptor obtained through histogram of oriented gradients, local binary pattern and Gabor filters. PCA- principal component analysis is used for dimensionality reduction of feature descriptor and k-nearest neighbors as classifier. The average emotion recognition accuracy achieved through method 1 and method 2 is 85.24% and 93.86% respectively. Effectiveness of both the techniques is compared on the basis of performance parameters such as accuracy, false positive rate, false negative rate and emotion recognition time. Emotion recognition has wide application areas so the work carried out can be applied for suitable application development.
Keywords: Emotion Recognition, Facial Expressions, Local Directional Number Histogram of Oriented Gradients, Hybrid Features.
Scope of the Article: Pattern Recognition