Cross-Cultural Emotion Analysis A Clustering Approach

It is stated that the various studies on the facial expression representations are varies from one culture to another culture and these facial expression are not universal. There exist different facial expression recognition (FER) systems that are suitable for the small size of datasets and generate good results. If same these experiments are perform on the different cultural datasets or large size of data sets the efficiency decrease radically. To maintain accuracy for a large datasets and different cultures, we used a novel hybrid clustering approach that is combination of the k-means and self-organizing map (SOM) clustering. In this research we used the local Binary Pattern and Histogram of Oriented Gradient for extraction of facial expression. We studied six common emotions such as anger, disgust, fear, happiness, sadness, and surprise of different cultures. In this research we used the following datasets for experiment to get our results. JAFFE, KDEF, TFEID, RadBoud, CK+, which originate from different cultures such as Japanese, Taiwanese, Moroccans, Caucasians, Afro-American, Euro-American, Asians, and Europeans. By applying this clustering approach we got 85% average Accuracy.