Facial Expressions Based Emotion Recognition through Feature Fusion Approach
Abstract— Human computer interaction is need of the time and has wide application range. Automatic emotion recognition is one of the important tasks which can be implemented through HCI systems. Facial expressions based non-verbal approach of emotion recognition is experimented for the work. Extracted features are fusion of features obtained through LDN and PCA. Extracted features are provided to feed forward neural network for emotion classification. The experimentation is carried out for CK+ and JAFFE database. The emotion recognition accuracy obtained for CK+ database is 82.07% and for JAFFE database is 84.22%. FAR - false acceptance rate and FRR - false rejection rate obtained for CK+ database is 4.18% and 17.92% respectively. FAR and FRR obtained for JAFFE database is 3.60% and 15.77% respectively.
Index Terms— local directional number, facial expression analysis, principle component analysis.
Sinhgad College of Engineering, INDIA
Zeal College of Engineering and Research, INDIA
Cite: Bharati Dixit, Arun Gaikwad, "Facial Expressions Based Emotion Recognition through Feature Fusion Approach," Proceedings of 2018 the 8th International Workshop on Computer Science and Engineering, pp. 258-263, Bangkok, 28-30 June, 2018.