Facial Beauty Analysis Model Based On Beauty Indicator

Along thousands of years, people are constantly looking for the good things. Themost closely related to human beings is the person s appearance beauty. The research offacial beauty has developed from cognitive psychology for facial beauty to computer orartificial intelligence. According to results of these researches, we can get theconclusion: the facial biological features can be extracted for the study of facial beauty.In this thesis, the relationship between the facial beauty indicator and facial biologicalfeatures is found. Through the extraction of facial point feature, the geometry featureand the proportion feature, establishing the objective facial beauty indicator analysismodel. The main research contents including:We proposed a novel method about facial beauty feature extraction and featurenormalization. How to define effective aesthetic features is fundamental for buildingfacial beauty analysis model. How to translate the understanding of human vision forbiological structure to automatic detection and analysis by computer is the first problemneed to be solved. According to feature of facial image, a set of facial beauty analysisfeatures is defined and extracted, including point feature, geometry feature andproportion feature. Facial beauty features are normalized according to some rules inorder to keep stability of them.Additionally, we also define the metric of subjective facial beauty evaluation. Inthe research of facial beauty indicator, metric of subjective facial beauty evaluation isnecessary to be defined. On the one hand, it can guide to establish the criteria ofobjective facial beauty indicator analysis model. On the other hand, it can verify thevalidity of objective facial beauty indicator analysis model. In this paper, we propose amethod to rate beauty indicator. First, subjective facial beauty evaluation is establishing,and then a large number of volunteers are invited for subjective evaluation which need acertain amount of facial beauty images. The scoring results are statistically analyzed.The suitable facial beauty indicator level is yielded by defining different beautyindicator and statistical results.Objective facial beauty indicator analysis model is built. According to subjectivefacial beauty evaluation, we build objective facial beauty indicator analysis model basedon improved K-nearest neighbor. In this model, classify by fusing different facial beautyfeature. The effectiveness of objective facial beauty indicator analysis model is tested according to subjective facial beauty evaluation.Through the research of facial beauty indicator, this paper set up an automaticfacial beauty indicator score system. This will make an influence on facial beauty andgive some useful enlightenment on Medical plastic in the future.