A Study on the Design of Facial Recognition and Classification Model Using Eye and Eyebrow Features: Focusing on Objectivity in Providing Physiognomic Information

With the rising societal interest and discussions around physiognomy, there is a growing emphasis on improving first impressions through the representation of facial images from a physiognomic perspective. Concurrently, there is an increasing demand for technologies that can automatically recognize and classify faces in photographs and videos. The purpose of this study is to design a facial recognition and classification model by utilizing the features of the eyes and eyebrows based on physiognomy. The study method involves analyzing physiognomic characteristics, identifying the names and positions of various parts of the eyes and eyebrows, and representing these features with coordinate values for mathematical verification. Features were measured on facial photographs using the ruler tool in Adobe Photoshop and then converted into ratios. Additionally, design aspects related to the shapes of the eyes and eyebrows were informed by the opinions of five makeup experts. The validation based on eyebrow shapes indicated the possibility of objectively quantifying faces with wide or narrow brows, as well as determining facial features based on eyebrow length and thickness. To validate based on the distance between the eyes, an analysis was conducted to distinguish faces with wide-set eyes from those with close-set eyes. Shapes between the eyes and eyebrows were classified and the results confirmed the feasibility of mathematical calculations for the distance between them. Based on these findings, the significance of this study is in presenting the feasibility of establishing criteria based on the shapes of eyes and eyebrows through geometric principles and mathematical equations. Furthermore, the objective facial analysis methods validated in this study could potentially enhance facial recognition systems and contribute to accessing facial analysis and physiognomic information in online users.