Feature selection for facial emotion recognition based on genetic algorithm
Hadjer Boubenna, Dohoon Lee · 2016
Facial emotion recognition is one of the most important subjects in image processing and computer vision fields. Through facial emotions the interaction human-machine can get more natural. In order to improve the accuracy we argue that feature selection is an important issue in facial emotion classification. We demonstrate that the error rate can be significantly reduced by removing some features that encode unimportant information from the image representation of faces. In this paper we propose genetic algorithm for feature selection. First the feature vector is extracted by using pyramid histogram of oriented gradient (PHOG) and then genetic algorithm (GA) is used to select a subset of features from the low-dimensional representation by removing certain values that seem to encode unimportant information about facial emotion. Finally, linear discriminant analysis (LDA) classifier is used to perform the classification. The results show that using GA as feature selector has significantly increased the accuracy. Compared to different approaches based on the well known dimensionality reduction technique namely principal component analysis (PCA) our approach leads to higher accuracy rate. The accuracy overall was 99.33%.