Analysis And Research Of Facial Attractiveness Based On Combined Geometric Features

In the field of genetic psychology and cognitive psychology,facial attractiveness has always been an important proposition,and the research results related to it can be used as an important scientific basis for human face evolution and human evolution,which has important scientific significance.Facial attractiveness using computer big data-assisted analysis belongs to the cross-research of Information and Psychology,which can solve the problems of strong subjectivity and insufficient data of traditional scoring in traditional research methods.In the research related to face attractiveness,geometric features are important factors affecting face attractiveness.In order to more comprehensively measure the relationship between geometric features and facial attractiveness,we will extract more detailed geometric features by combining facial features and facial distance ratio to predict facial attractiveness,so as to improve the computer's performance in predicting facial attractiveness scores.The specific research contents are as follows:(1)According to the shapes and area of the facial features,a method for constructing geometric models of eyebrows and eyes was proposed,and the geometric features such as length,area,average width,bending degree of eyebrow,size of eye and iris,and proportion of iris to whole eye were obtained.The STASM model was used to mark the 77 feature points of the face contour,and the ratio of the horizontal width to the vertical length of the face was obtained.(2)Statistical analysis of geometric feature parameters,analysis of eyebrow and eye shape matching in different attractive score intervals,and analysis of the relationship between face shape and eyebrow shape by combining the facial proportion and geometric features of eyebrows were performed respectively.Experimental results show that there is a close relationship between the shape features of the eyes,eyebrows and the facial proportions in different facial attractive levels.(3)Combining geometric features such as facial features and face distance ratios,the 77 feature points,21-dimensional distance features,triangle area features,and scale feature sets of the face were extracted for predicting the attractiveness of the face.The face attractive prediction performance of each geometric feature under four machine learning algorithms is compared in the self-collection dataset and SCUT-FBP dataset.The experimental results show that the geometric features proposed in this paper are reasonable and effective for characterizing the attractiveness of the face,and the combined feature set performs better in predicting performance.Under the SCUT-FBP dataset,the maximum Pearson coefficient of the combined feature is up to 0.904.