Study On Facial Beauty Classification

Beauty is widely perceived by the human conception, and facial beauty is the most common perception of objects in human social activities. Humans investigate the concept of beauty more than a few thousand years, not only to promote continuous improvements of the aesthetic consciousness of people, but also promote the rapid development of society. However, it is still challenging to define what is a beautiful face? Recently, numerous studies in cognitive psychology found what a beautiful face was and there is a high degree of consistency over different races, ages and genders. In our work, facial beauty classification methods based on image processing and machine learning techniques have been proposed. We adopt some quantitative methods to extract facial beauty features and beauty classification, with aims to seek some guiding significance to be used in clinical cosmetic, orthopedic and prosthetic in real worlds.While the research on facial beauty analysis via image processing and machine learning schemes is relatively small reported in the literature, it has begun to attract the attention of many researchers in recent years. By investigating the research problems on facial beauty analysis, we are like to give some objective and quantifiable descriptions. The main contributions of our work include the following four aspects:(1) A large scale face image dataset containing both male and female are collected to establish for experimental study. These face images are mainly drawled from some popular social networking sites.(2) To describe effective features to describe facial beauty information, the local binary pattern (LBP) descriptor and sub-block LBP (denoted by Block-LBP) are employed to extract the texture features, respectively.(3) The scale invariant feature transform (SIFT) descriptor with spatial pyramid matching (SPM) strategy is employed to extract facial feature points, and also used to represent facial beauty information.(4) Given a face image, we use the K-Nearest Neighbor (KNN) classifier and Support Vector Machine (SVM) classifier to perform the beauty classification task, respectively. Lastly, numerous discussions of experimental results are reported in detail.Experimental results show that our proposed facial beauty classification approaches are effective to assess human facial beauty. In addition, we draw some important conclusions.