Quantitative Analysis of Facial Symmetry Among Different Expressions
Ke Wu, Weiyang Chen · 2021
Facial expression recognition technology has gradually become a research hotspot. Facial symmetry is an important facial feature. This paper quantitatively analyzes the differences of various facial symmetry indexes among different facial expressions. Firstly, image preprocessing standardizes the images and finds out the landmarks. Then five asymmetric features of the angle of key landmarks, regional asymmetry, centroid, singular value, and structural dissimilarity are extracted from the facial images. The experimental results show the following conclusions. The asymmetric features of happy are more aggregated than other expressions. The mild asymmetry accounted for the largest proportion of happy. The moderate asymmetry accounted for the largest proportion of most expressions. The severe asymmetry of all expressions accounted for the least.