A Study For Facial Beauty Attractiveness Prediction Based On Deep Autoencoder

Beauty is a common essential attribute of objective things which causes people’s sense of beauty. In the field of art research, beauty is an abstract concept that people often encounter. It is a concept of the human senses, belongs to perceived range. The object that people often encounter in practice of human life is the face beauty. Facial beauty is not only able to give people pleasure in the mind, but also can cultivate people’s sentiments. What is beauty? Whether "Beauty Code" objectively exist in reality? Although we can easily perceive facial beauty attractiveness by own organs, but very difficultly define facial beauty attractiveness through precise manner. Cognitive psychologists did a lot of research experiments to explore the wonders of facial beauty attractiveness and obtained an conclusion:Regardless of race, culture, gender and age, there is a high consistency for what face is beautiful among people.In everyday life, we truly feel the beauty and attractiveness which unavoidably significantly impact on people. Because of this, many scientists and philosophers devoted their energy into beautiful attractiveness study, thereby advancing the research continue to move forward. In fact, in the computer and information processing disciplines, the use of image processing, image analysis, artificial intelligence and machine learning methods for objectively evaluating facial beauty research reports are not many, but the most recent years, many researchers have begun to pay attention on it again. Also, in the field of artificial intelligence, deep learning network has been a major breakthrough, you can learn apparent facial features from face image. Inspired, this article will apply t to learn face beautiful features, and predict facial beauty attractiveness with machine learning method. This paper focuses on facial beauty features extraction method, appearance features, Deep Autoencoder and machine learning to carry out research work.The main work includes:(1)To systematically review the theoretical basis of facial beauty attractiveness research, facial beauty attractiveness objective existence theory and facial beauty attractiveness learnability theory.(2)To systematically review facial beauty attractiveness prediction model based on the geometric features studied by Predecessor, and point out the deficiency of facial beauty geometric features, thus lead to apparent features, finally discuss these in this paper.(3)To Review facial beauty Eigenface and Gabor features, and systematically review support vector machine learning method.(4)To put forward facial beauty attractiveness prediction model based on Deep Autoencoder, and further study Deep Autoencoder, finally the experimental results show that encoder network of Deep Autoencoder is used as facial beauty features extraction, and combined with the support vector machine (SVM) to predict is feasible.In recent years, the research on facial beauty is a new study which beginning to appear in image processing and pattern recognition and flourished. Research work of this paper is to use the image information processing technology, deep learning networks, and support vector machine methods did a very meaningful attempt on facial beauty and analyzes attractive IntelliSense. Facial beauty attractiveness research has just started, and not yet mature, there are still a lot of research room to grow. Looking forward to the development of computer science and image information processing technology can provide powerful technology and means, as a strong backing to the face beautiful attractive research and better dig the beauty’s inner essence, enable the computer to like people perceive beauty and possess intelligent perception.