Facial Expressions: A Cross‐Cultural Study
Chandrani Saha, Washef Ahmed, Soma Mitra, Debasis Mazumdar, Sushmita Mitra · 2015
This chapter presents a cross-cultural study on facial expressions. It computationally shows that the occurrence of Action Units (AUs) also possess inter-culture variations. The chapter describes the extraction of facial regions bearing prominent muscle movement for different emotional facial expressions. These facial regions are then mapped with Ekman's AUs. The chapter discusses the methodology to extract AUs from image sequences of basic expressions. The image sequences are obtained by splitting videos into successive frames. The chapter explores the cultural variability in the occurrence of different AUs in six basic facial expressions. It also describes the generation of rule base using decision tree for facial expression classification, the performance of which is evaluated in the context of cultural variability. Finally, it is demonstrated that the performance of the rule-based classifier improves when the rule base becomes culture specific.