Facial emotion recognition is very important for social communication. Whereby through the years it has done many studies and researches about automatic emotion recognition. Generally, the facial emotion recognition systems are composed of the pre-processing phase, the features extraction phase and the classification phase. This article proposes a method for automatic facial emotion recognition in digital images. This method uses histogram equalization to improve special lighting conditions during the pre-processing of images, two-dimensional discrete wavelet transform and PCA algorithms are used to extract and reduce features. Finally it uses a support vector machine linear to predict emotions. The experiments were made using the JAFFE database and CK+. The method achieves more than 93% for average accuracy and it recognizes better the following emotions: Happiness, neutral and surprise. For comparisons, two kinds of classifiers were adopted: Support vector machine and Convolutional Neural Network. Using this last classifier we achieve more than 98% for average accuracy.
|Title of host publication||Proceedings of the 2020 IEEE Engineering International Research Conference, EIRCON 2020|
|Publisher||Institute of Electrical and Electronics Engineers Inc.|
|State||Published - 21 Oct 2020|
|Event||2020 IEEE Engineering International Research Conference, EIRCON 2020 - Lima, Peru|
Duration: 21 Oct 2020 → 23 Oct 2020
|Name||Proceedings of the 2020 IEEE Engineering International Research Conference, EIRCON 2020|
|Conference||2020 IEEE Engineering International Research Conference, EIRCON 2020|
|Period||21/10/20 → 23/10/20|
Bibliographical notePublisher Copyright:
© 2020 IEEE.
- Facial Emotion Recognition
- Facial Expression Classification
- Features Extraction
- Images Pre-processing