Analysis of Eyebrow Motion for Micro-Expression Recognition
Sultan Almalki, Usman Saeed, Mohammed Saeed Alkatheiri · 2022
Facial expression recognition is a field of emotion recognition that has received considerable attention over the last decades. However, most of the research has been focused on the recognition of macro-expressions that are generally posed and easy to imitate. On the other hand, micro-expressions are much subtler and localized therefore by definition harder to replicate, thus making them valuable for medical diagnosis and criminal investigations. Normally, micro-expressions are detected and recognized by studying the entire face. We believe that the eyebrows contain the most relevant information pertaining to micro-expression recognition. Therefore, in this paper, we propose to recognize facial micro-expressions based solely on the motion of the eyebrow. The proposed method starts with the preprocessing and cropping of image sequences based on the eye area. Next, feature extraction is carried out using Histogram of Oriented Optical Flow (HOOF), and Local Binary Patterns on Three Orthogonal Planes (LBP-TOP). Validation of our methodology is carried out on the SMIC-HS and CASME II databases. We achieved promising recognition accuracy of 54% with magnification and 64% without magnification on the SMIC-HS database. In the case of CASME II database we achieved an accuracy of 57% with magnification, and 59% without magnification.