Facial Beautification Method Based on Age Evolution  

作　　者：CHEN Yan DING Shou-hong HU Gan-le MA Li-zhuang 

机构地区：[1]Department of Computer Science and Engineering, Shanghai Jiao Tong University

出　　处：《Computer Aided Drafting,Design and Manufacturing》2013年第4期7-13,共7页计算机辅助绘图设计与制造（英文版）

基　　金：Supported by NSFC Key Project(No.61133009)

摘　　要：This paper proposes a new facial beautification method using facial rejuvenation based on the age evolution. Traditional facial beautification methods only focus on the color of skin and deformation and do the transformation based on an experimental standard of beauty. Our method achieves the beauty effect by making facial image looks younger, which is different from traditional methods and is more reasonable than them. Firstly, we decompose the image into different layers and get a detail layer. Secondly, we get an age-related parameter： the standard deviation of the Gaussian distribution that the detail layer follows, and the support vector machine （SVM） regression is used to fit a function about the age and the standard deviation. Thirdly, we use this function to estimate the age of input image and generate a new detail layer with a new standard deviation which is calculated by decreasing the age. Lastly, we combine the original layers and the new detail layer to get a new face image. Experimental results show that this algo- rithm can make facial image become more beautiful by facial rejuvenation. The proposed method opens up a new way about facial beautification, and there are great potentials for applications.This paper proposes a new facial beautification method using facial rejuvenation based on the age evolution. Traditional facial beautification methods only focus on the color of skin and deformation and do the transformation based on an experimental standard of beauty. Our method achieves the beauty effect by making facial image looks younger, which is different from traditional methods and is more reasonable than them. Firstly, we decompose the image into different layers and get a detail layer. Secondly, we get an age-related parameter： the standard deviation of the Gaussian distribution that the detail layer follows, and the support vector machine （SVM） regression is used to fit a function about the age and the standard deviation. Thirdly, we use this function to estimate the age of input image and generate a new detail layer with a new standard deviation which is calculated by decreasing the age. Lastly, we combine the original layers and the new detail layer to get a new face image. Experimental results show that this algo- rithm can make facial image become more beautiful by facial rejuvenation. The proposed method opens up a new way about facial beautification, and there are great potentials for applications.

关 键 词：facial beautification method age evolution support vector machine 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TS974.1[自动化与计算机技术—计算机科学与技术]